Datasets:
Commit ·
bc36326
0
Parent(s):
Initial public release
Browse files- .gitattributes +60 -0
- LICENSE +202 -0
- README.md +93 -0
- SHA256SUMS +23 -0
- data/rate_annotations/test.jsonl +0 -0
- prompts/README.md +38 -0
- prompts/additional/orbit_sc_sbs.json +28 -0
- prompts/additional/orbit_sc_sbs.md +428 -0
- prompts/additional/rate_full_with_examples_and_coverage_k2.json +28 -0
- prompts/additional/rate_full_with_examples_and_coverage_k2.md +333 -0
- prompts/esa_ae.json +8 -0
- prompts/esa_ae.md +53 -0
- prompts/mqm_ae.json +8 -0
- prompts/mqm_ae.md +97 -0
- prompts/orbit_mc_fuser.json +28 -0
- prompts/orbit_mc_fuser.md +477 -0
- prompts/orbit_sc.json +28 -0
- prompts/orbit_sc.md +246 -0
- prompts/prompts.jsonl +0 -0
- prompts/rate_brief.json +8 -0
- prompts/rate_brief.md +59 -0
- prompts/rate_full.json +12 -0
- prompts/rate_full.md +99 -0
- prompts/rate_full_with_examples.json +8 -0
- prompts/rate_full_with_examples.md +146 -0
.gitattributes
ADDED
|
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
*.7z filter=lfs diff=lfs merge=lfs -text
|
| 2 |
+
*.arrow filter=lfs diff=lfs merge=lfs -text
|
| 3 |
+
*.avro filter=lfs diff=lfs merge=lfs -text
|
| 4 |
+
*.bin filter=lfs diff=lfs merge=lfs -text
|
| 5 |
+
*.bz2 filter=lfs diff=lfs merge=lfs -text
|
| 6 |
+
*.ckpt filter=lfs diff=lfs merge=lfs -text
|
| 7 |
+
*.ftz filter=lfs diff=lfs merge=lfs -text
|
| 8 |
+
*.gz filter=lfs diff=lfs merge=lfs -text
|
| 9 |
+
*.h5 filter=lfs diff=lfs merge=lfs -text
|
| 10 |
+
*.joblib filter=lfs diff=lfs merge=lfs -text
|
| 11 |
+
*.lfs.* filter=lfs diff=lfs merge=lfs -text
|
| 12 |
+
*.lz4 filter=lfs diff=lfs merge=lfs -text
|
| 13 |
+
*.mds filter=lfs diff=lfs merge=lfs -text
|
| 14 |
+
*.mlmodel filter=lfs diff=lfs merge=lfs -text
|
| 15 |
+
*.model filter=lfs diff=lfs merge=lfs -text
|
| 16 |
+
*.msgpack filter=lfs diff=lfs merge=lfs -text
|
| 17 |
+
*.npy filter=lfs diff=lfs merge=lfs -text
|
| 18 |
+
*.npz filter=lfs diff=lfs merge=lfs -text
|
| 19 |
+
*.onnx filter=lfs diff=lfs merge=lfs -text
|
| 20 |
+
*.ot filter=lfs diff=lfs merge=lfs -text
|
| 21 |
+
*.parquet filter=lfs diff=lfs merge=lfs -text
|
| 22 |
+
*.pb filter=lfs diff=lfs merge=lfs -text
|
| 23 |
+
*.pickle filter=lfs diff=lfs merge=lfs -text
|
| 24 |
+
*.pkl filter=lfs diff=lfs merge=lfs -text
|
| 25 |
+
*.pt filter=lfs diff=lfs merge=lfs -text
|
| 26 |
+
*.pth filter=lfs diff=lfs merge=lfs -text
|
| 27 |
+
*.rar filter=lfs diff=lfs merge=lfs -text
|
| 28 |
+
*.safetensors filter=lfs diff=lfs merge=lfs -text
|
| 29 |
+
saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
| 30 |
+
*.tar.* filter=lfs diff=lfs merge=lfs -text
|
| 31 |
+
*.tar filter=lfs diff=lfs merge=lfs -text
|
| 32 |
+
*.tflite filter=lfs diff=lfs merge=lfs -text
|
| 33 |
+
*.tgz filter=lfs diff=lfs merge=lfs -text
|
| 34 |
+
*.wasm filter=lfs diff=lfs merge=lfs -text
|
| 35 |
+
*.xz filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 37 |
+
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 38 |
+
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 39 |
+
# Audio files - uncompressed
|
| 40 |
+
*.pcm filter=lfs diff=lfs merge=lfs -text
|
| 41 |
+
*.sam filter=lfs diff=lfs merge=lfs -text
|
| 42 |
+
*.raw filter=lfs diff=lfs merge=lfs -text
|
| 43 |
+
# Audio files - compressed
|
| 44 |
+
*.aac filter=lfs diff=lfs merge=lfs -text
|
| 45 |
+
*.flac filter=lfs diff=lfs merge=lfs -text
|
| 46 |
+
*.mp3 filter=lfs diff=lfs merge=lfs -text
|
| 47 |
+
*.ogg filter=lfs diff=lfs merge=lfs -text
|
| 48 |
+
*.wav filter=lfs diff=lfs merge=lfs -text
|
| 49 |
+
# Image files - uncompressed
|
| 50 |
+
*.bmp filter=lfs diff=lfs merge=lfs -text
|
| 51 |
+
*.gif filter=lfs diff=lfs merge=lfs -text
|
| 52 |
+
*.png filter=lfs diff=lfs merge=lfs -text
|
| 53 |
+
*.tiff filter=lfs diff=lfs merge=lfs -text
|
| 54 |
+
# Image files - compressed
|
| 55 |
+
*.jpg filter=lfs diff=lfs merge=lfs -text
|
| 56 |
+
*.jpeg filter=lfs diff=lfs merge=lfs -text
|
| 57 |
+
*.webp filter=lfs diff=lfs merge=lfs -text
|
| 58 |
+
# Video files - compressed
|
| 59 |
+
*.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 60 |
+
*.webm filter=lfs diff=lfs merge=lfs -text
|
LICENSE
ADDED
|
@@ -0,0 +1,202 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
Apache License
|
| 3 |
+
Version 2.0, January 2004
|
| 4 |
+
http://www.apache.org/licenses/
|
| 5 |
+
|
| 6 |
+
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
|
| 7 |
+
|
| 8 |
+
1. Definitions.
|
| 9 |
+
|
| 10 |
+
"License" shall mean the terms and conditions for use, reproduction,
|
| 11 |
+
and distribution as defined by Sections 1 through 9 of this document.
|
| 12 |
+
|
| 13 |
+
"Licensor" shall mean the copyright owner or entity authorized by
|
| 14 |
+
the copyright owner that is granting the License.
|
| 15 |
+
|
| 16 |
+
"Legal Entity" shall mean the union of the acting entity and all
|
| 17 |
+
other entities that control, are controlled by, or are under common
|
| 18 |
+
control with that entity. For the purposes of this definition,
|
| 19 |
+
"control" means (i) the power, direct or indirect, to cause the
|
| 20 |
+
direction or management of such entity, whether by contract or
|
| 21 |
+
otherwise, or (ii) ownership of fifty percent (50%) or more of the
|
| 22 |
+
outstanding shares, or (iii) beneficial ownership of such entity.
|
| 23 |
+
|
| 24 |
+
"You" (or "Your") shall mean an individual or Legal Entity
|
| 25 |
+
exercising permissions granted by this License.
|
| 26 |
+
|
| 27 |
+
"Source" form shall mean the preferred form for making modifications,
|
| 28 |
+
including but not limited to software source code, documentation
|
| 29 |
+
source, and configuration files.
|
| 30 |
+
|
| 31 |
+
"Object" form shall mean any form resulting from mechanical
|
| 32 |
+
transformation or translation of a Source form, including but
|
| 33 |
+
not limited to compiled object code, generated documentation,
|
| 34 |
+
and conversions to other media types.
|
| 35 |
+
|
| 36 |
+
"Work" shall mean the work of authorship, whether in Source or
|
| 37 |
+
Object form, made available under the License, as indicated by a
|
| 38 |
+
copyright notice that is included in or attached to the work
|
| 39 |
+
(an example is provided in the Appendix below).
|
| 40 |
+
|
| 41 |
+
"Derivative Works" shall mean any work, whether in Source or Object
|
| 42 |
+
form, that is based on (or derived from) the Work and for which the
|
| 43 |
+
editorial revisions, annotations, elaborations, or other modifications
|
| 44 |
+
represent, as a whole, an original work of authorship. For the purposes
|
| 45 |
+
of this License, Derivative Works shall not include works that remain
|
| 46 |
+
separable from, or merely link (or bind by name) to the interfaces of,
|
| 47 |
+
the Work and Derivative Works thereof.
|
| 48 |
+
|
| 49 |
+
"Contribution" shall mean any work of authorship, including
|
| 50 |
+
the original version of the Work and any modifications or additions
|
| 51 |
+
to that Work or Derivative Works thereof, that is intentionally
|
| 52 |
+
submitted to Licensor for inclusion in the Work by the copyright owner
|
| 53 |
+
or by an individual or Legal Entity authorized to submit on behalf of
|
| 54 |
+
the copyright owner. For the purposes of this definition, "submitted"
|
| 55 |
+
means any form of electronic, verbal, or written communication sent
|
| 56 |
+
to the Licensor or its representatives, including but not limited to
|
| 57 |
+
communication on electronic mailing lists, source code control systems,
|
| 58 |
+
and issue tracking systems that are managed by, or on behalf of, the
|
| 59 |
+
Licensor for the purpose of discussing and improving the Work, but
|
| 60 |
+
excluding communication that is conspicuously marked or otherwise
|
| 61 |
+
designated in writing by the copyright owner as "Not a Contribution."
|
| 62 |
+
|
| 63 |
+
"Contributor" shall mean Licensor and any individual or Legal Entity
|
| 64 |
+
on behalf of whom a Contribution has been received by Licensor and
|
| 65 |
+
subsequently incorporated within the Work.
|
| 66 |
+
|
| 67 |
+
2. Grant of Copyright License. Subject to the terms and conditions of
|
| 68 |
+
this License, each Contributor hereby grants to You a perpetual,
|
| 69 |
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
| 70 |
+
copyright license to reproduce, prepare Derivative Works of,
|
| 71 |
+
publicly display, publicly perform, sublicense, and distribute the
|
| 72 |
+
Work and such Derivative Works in Source or Object form.
|
| 73 |
+
|
| 74 |
+
3. Grant of Patent License. Subject to the terms and conditions of
|
| 75 |
+
this License, each Contributor hereby grants to You a perpetual,
|
| 76 |
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
| 77 |
+
(except as stated in this section) patent license to make, have made,
|
| 78 |
+
use, offer to sell, sell, import, and otherwise transfer the Work,
|
| 79 |
+
where such license applies only to those patent claims licensable
|
| 80 |
+
by such Contributor that are necessarily infringed by their
|
| 81 |
+
Contribution(s) alone or by combination of their Contribution(s)
|
| 82 |
+
with the Work to which such Contribution(s) was submitted. If You
|
| 83 |
+
institute patent litigation against any entity (including a
|
| 84 |
+
cross-claim or counterclaim in a lawsuit) alleging that the Work
|
| 85 |
+
or a Contribution incorporated within the Work constitutes direct
|
| 86 |
+
or contributory patent infringement, then any patent licenses
|
| 87 |
+
granted to You under this License for that Work shall terminate
|
| 88 |
+
as of the date such litigation is filed.
|
| 89 |
+
|
| 90 |
+
4. Redistribution. You may reproduce and distribute copies of the
|
| 91 |
+
Work or Derivative Works thereof in any medium, with or without
|
| 92 |
+
modifications, and in Source or Object form, provided that You
|
| 93 |
+
meet the following conditions:
|
| 94 |
+
|
| 95 |
+
(a) You must give any other recipients of the Work or
|
| 96 |
+
Derivative Works a copy of this License; and
|
| 97 |
+
|
| 98 |
+
(b) You must cause any modified files to carry prominent notices
|
| 99 |
+
stating that You changed the files; and
|
| 100 |
+
|
| 101 |
+
(c) You must retain, in the Source form of any Derivative Works
|
| 102 |
+
that You distribute, all copyright, patent, trademark, and
|
| 103 |
+
attribution notices from the Source form of the Work,
|
| 104 |
+
excluding those notices that do not pertain to any part of
|
| 105 |
+
the Derivative Works; and
|
| 106 |
+
|
| 107 |
+
(d) If the Work includes a "NOTICE" text file as part of its
|
| 108 |
+
distribution, then any Derivative Works that You distribute must
|
| 109 |
+
include a readable copy of the attribution notices contained
|
| 110 |
+
within such NOTICE file, excluding those notices that do not
|
| 111 |
+
pertain to any part of the Derivative Works, in at least one
|
| 112 |
+
of the following places: within a NOTICE text file distributed
|
| 113 |
+
as part of the Derivative Works; within the Source form or
|
| 114 |
+
documentation, if provided along with the Derivative Works; or,
|
| 115 |
+
within a display generated by the Derivative Works, if and
|
| 116 |
+
wherever such third-party notices normally appear. The contents
|
| 117 |
+
of the NOTICE file are for informational purposes only and
|
| 118 |
+
do not modify the License. You may add Your own attribution
|
| 119 |
+
notices within Derivative Works that You distribute, alongside
|
| 120 |
+
or as an addendum to the NOTICE text from the Work, provided
|
| 121 |
+
that such additional attribution notices cannot be construed
|
| 122 |
+
as modifying the License.
|
| 123 |
+
|
| 124 |
+
You may add Your own copyright statement to Your modifications and
|
| 125 |
+
may provide additional or different license terms and conditions
|
| 126 |
+
for use, reproduction, or distribution of Your modifications, or
|
| 127 |
+
for any such Derivative Works as a whole, provided Your use,
|
| 128 |
+
reproduction, and distribution of the Work otherwise complies with
|
| 129 |
+
the conditions stated in this License.
|
| 130 |
+
|
| 131 |
+
5. Submission of Contributions. Unless You explicitly state otherwise,
|
| 132 |
+
any Contribution intentionally submitted for inclusion in the Work
|
| 133 |
+
by You to the Licensor shall be under the terms and conditions of
|
| 134 |
+
this License, without any additional terms or conditions.
|
| 135 |
+
Notwithstanding the above, nothing herein shall supersede or modify
|
| 136 |
+
the terms of any separate license agreement you may have executed
|
| 137 |
+
with Licensor regarding such Contributions.
|
| 138 |
+
|
| 139 |
+
6. Trademarks. This License does not grant permission to use the trade
|
| 140 |
+
names, trademarks, service marks, or product names of the Licensor,
|
| 141 |
+
except as required for reasonable and customary use in describing the
|
| 142 |
+
origin of the Work and reproducing the content of the NOTICE file.
|
| 143 |
+
|
| 144 |
+
7. Disclaimer of Warranty. Unless required by applicable law or
|
| 145 |
+
agreed to in writing, Licensor provides the Work (and each
|
| 146 |
+
Contributor provides its Contributions) on an "AS IS" BASIS,
|
| 147 |
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
|
| 148 |
+
implied, including, without limitation, any warranties or conditions
|
| 149 |
+
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
|
| 150 |
+
PARTICULAR PURPOSE. You are solely responsible for determining the
|
| 151 |
+
appropriateness of using or redistributing the Work and assume any
|
| 152 |
+
risks associated with Your exercise of permissions under this License.
|
| 153 |
+
|
| 154 |
+
8. Limitation of Liability. In no event and under no legal theory,
|
| 155 |
+
whether in tort (including negligence), contract, or otherwise,
|
| 156 |
+
unless required by applicable law (such as deliberate and grossly
|
| 157 |
+
negligent acts) or agreed to in writing, shall any Contributor be
|
| 158 |
+
liable to You for damages, including any direct, indirect, special,
|
| 159 |
+
incidental, or consequential damages of any character arising as a
|
| 160 |
+
result of this License or out of the use or inability to use the
|
| 161 |
+
Work (including but not limited to damages for loss of goodwill,
|
| 162 |
+
work stoppage, computer failure or malfunction, or any and all
|
| 163 |
+
other commercial damages or losses), even if such Contributor
|
| 164 |
+
has been advised of the possibility of such damages.
|
| 165 |
+
|
| 166 |
+
9. Accepting Warranty or Additional Liability. While redistributing
|
| 167 |
+
the Work or Derivative Works thereof, You may choose to offer,
|
| 168 |
+
and charge a fee for, acceptance of support, warranty, indemnity,
|
| 169 |
+
or other liability obligations and/or rights consistent with this
|
| 170 |
+
License. However, in accepting such obligations, You may act only
|
| 171 |
+
on Your own behalf and on Your sole responsibility, not on behalf
|
| 172 |
+
of any other Contributor, and only if You agree to indemnify,
|
| 173 |
+
defend, and hold each Contributor harmless for any liability
|
| 174 |
+
incurred by, or claims asserted against, such Contributor by reason
|
| 175 |
+
of your accepting any such warranty or additional liability.
|
| 176 |
+
|
| 177 |
+
END OF TERMS AND CONDITIONS
|
| 178 |
+
|
| 179 |
+
APPENDIX: How to apply the Apache License to your work.
|
| 180 |
+
|
| 181 |
+
To apply the Apache License to your work, attach the following
|
| 182 |
+
boilerplate notice, with the fields enclosed by brackets "[]"
|
| 183 |
+
replaced with your own identifying information. (Don't include
|
| 184 |
+
the brackets!) The text should be enclosed in the appropriate
|
| 185 |
+
comment syntax for the file format. We also recommend that a
|
| 186 |
+
file or class name and description of purpose be included on the
|
| 187 |
+
same "printed page" as the copyright notice for easier
|
| 188 |
+
identification within third-party archives.
|
| 189 |
+
|
| 190 |
+
Copyright [yyyy] [name of copyright owner]
|
| 191 |
+
|
| 192 |
+
Licensed under the Apache License, Version 2.0 (the "License");
|
| 193 |
+
you may not use this file except in compliance with the License.
|
| 194 |
+
You may obtain a copy of the License at
|
| 195 |
+
|
| 196 |
+
http://www.apache.org/licenses/LICENSE-2.0
|
| 197 |
+
|
| 198 |
+
Unless required by applicable law or agreed to in writing, software
|
| 199 |
+
distributed under the License is distributed on an "AS IS" BASIS,
|
| 200 |
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 201 |
+
See the License for the specific language governing permissions and
|
| 202 |
+
limitations under the License.
|
README.md
ADDED
|
@@ -0,0 +1,93 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
pretty_name: "ORBIT-MT-Eval: Expert Annotations and Prompts for Machine Translation Meta-Evaluation"
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
- ru
|
| 6 |
+
license: apache-2.0
|
| 7 |
+
task_categories:
|
| 8 |
+
- translation
|
| 9 |
+
tags:
|
| 10 |
+
- machine-translation
|
| 11 |
+
- evaluation
|
| 12 |
+
- error-analysis
|
| 13 |
+
- rate
|
| 14 |
+
size_categories:
|
| 15 |
+
- n<10K
|
| 16 |
+
configs:
|
| 17 |
+
- config_name: rate_annotations
|
| 18 |
+
data_files:
|
| 19 |
+
- split: test
|
| 20 |
+
path: data/rate_annotations/test.jsonl
|
| 21 |
+
---
|
| 22 |
+
|
| 23 |
+
# ORBIT-MT-Eval: Expert Annotations and Prompts for Machine Translation Meta-Evaluation
|
| 24 |
+
|
| 25 |
+
ORBIT-MT-Eval is an English-to-Russian reference dataset for machine translation meta-evaluation: comparing evaluator outputs against professional expert annotations. Each of the 600 source–translation segments has three independent expert annotations following the [RATE protocol](https://aclanthology.org/2025.findings-emnlp.1203/). In *Beyond Prompts: A Systematic Study of LLM-Based Machine Translation Evaluators*, two annotations serve as gold references and the third is scored against each reference and averaged to produce the Human_E agreement baseline. The release also includes the prompts used by ORBIT and the locally maintained prompting baselines. GEMBA and `unified-mqm-boosted-v5` are linked to their official upstream implementations.
|
| 26 |
+
|
| 27 |
+
## Dataset at a glance
|
| 28 |
+
|
| 29 |
+
| Property | Value |
|
| 30 |
+
|---|---|
|
| 31 |
+
| Language direction | English → Russian |
|
| 32 |
+
| Source segments | 300 |
|
| 33 |
+
| Source–translation segments | 600 (two translations for each source) |
|
| 34 |
+
| Professional expert annotators | 35 |
|
| 35 |
+
| Expert annotations | 3 independent annotations per segment from the expert pool, for 1,800 records total |
|
| 36 |
+
| Meta-evaluation protocol | 2 reference annotations + 1 Human_E agreement baseline |
|
| 37 |
+
| Error spans | 9,771 |
|
| 38 |
+
| Annotation scheme | RATE severity 1–5 with separate Accuracy and Fluency scores |
|
| 39 |
+
| Domains | Literary, news, social media, and speech |
|
| 40 |
+
|
| 41 |
+
`expert_1`, `expert_2`, and `expert_3` identify the three annotation slots within each segment. They are not person-level IDs and do not track individual annotators across the dataset.
|
| 42 |
+
|
| 43 |
+
## Files
|
| 44 |
+
|
| 45 |
+
- [`data/rate_annotations/test.jsonl`](data/rate_annotations/test.jsonl): the annotation records.
|
| 46 |
+
- [`prompts/README.md`](prompts/README.md): readable prompt pages with copy-ready JSON `messages` objects.
|
| 47 |
+
- [`prompts/prompts.jsonl`](prompts/prompts.jsonl): the machine-readable prompt collection.
|
| 48 |
+
|
| 49 |
+
## Load
|
| 50 |
+
|
| 51 |
+
```python
|
| 52 |
+
from datasets import load_dataset
|
| 53 |
+
|
| 54 |
+
dataset = load_dataset("foksly/orbit-mt-eval", "rate_annotations", split="test")
|
| 55 |
+
print(dataset[0])
|
| 56 |
+
```
|
| 57 |
+
|
| 58 |
+
## Paper results
|
| 59 |
+
|
| 60 |
+
| Evaluator | R | P | F1 | R4+ | P4+ | MQM |
|
| 61 |
+
|---|---:|---:|---:|---:|---:|---:|
|
| 62 |
+
| Human_E | 46.8 | 44.3 | 45.5 | 65.3 | 65.3 | 75.3 |
|
| 63 |
+
| Human_A-SbS | 39.3 | 47.1 | 42.8 | 55.7 | 63.2 | 68.0 |
|
| 64 |
+
| Human_A-PW | 31.8 | 46.4 | 37.7 | 49.8 | 57.1 | 65.0 |
|
| 65 |
+
| Human_WMT | 12.3 | 61.9 | 20.6 | 26.6 | 62.8 | 42.7 |
|
| 66 |
+
| GEMBA-MQM | 24.6 | 43.0 | 31.3 | 45.0 | 51.0 | 58.7 |
|
| 67 |
+
| GEMBA-ESA | 24.9 | 41.3 | 31.0 | 44.6 | 45.7 | 57.3 |
|
| 68 |
+
| MQM-AE | 24.0 | 51.8 | 32.8 | 45.4 | 54.5 | 60.0 |
|
| 69 |
+
| ESA-AE | 27.0 | 54.8 | 36.2 | 49.5 | 58.6 | 53.7 |
|
| 70 |
+
| UMB-v5 | 27.8 | 44.5 | 34.2 | 48.7 | 56.4 | 57.3 |
|
| 71 |
+
| xCOMET-XXL | 18.3 | 36.4 | 24.3 | 33.5 | 36.8 | 58.3 |
|
| 72 |
+
| ORBIT-SC | 51.5 | 46.2 | 48.7 | 74.7 | 63.5 | 67.0 |
|
| 73 |
+
| ORBIT-MC | **60.3** | 45.8 | **52.1** | **82.0** | **70.6** | **69.3** |
|
| 74 |
+
|
| 75 |
+
Values are on a 0–100 scale for the English→Russian evaluation set. R/P/F1 are micro-averaged span metrics excluding RATE severity 1, R4+/P4+ use severity ≥4, and MQM is pairwise ranking accuracy.
|
| 76 |
+
|
| 77 |
+
Prompting baselines and ORBIT-SC use GPT-5.4. ORBIT-MC fuses three evaluator backbones with an LLM fuser.
|
| 78 |
+
|
| 79 |
+
## Citation
|
| 80 |
+
|
| 81 |
+
```bibtex
|
| 82 |
+
@inproceedings{popov-etal-2026-beyond-prompts,
|
| 83 |
+
title = {Beyond Prompts: A Systematic Study of LLM-Based Machine Translation Evaluators},
|
| 84 |
+
author = {Popov, Dmitry and Bokhyan, Roman and Enikeeva, Ekaterina and
|
| 85 |
+
Mekhraliev, Artem and Vysotsky, Stepan and Karpachev, Nikolay},
|
| 86 |
+
booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2026},
|
| 87 |
+
year = {2026}
|
| 88 |
+
}
|
| 89 |
+
```
|
| 90 |
+
|
| 91 |
+
## License
|
| 92 |
+
|
| 93 |
+
The original RATE annotations and prompt material are released under the [Apache License 2.0](LICENSE). Source and translation material retain their applicable upstream terms.
|
SHA256SUMS
ADDED
|
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
cfc7749b96f63bd31c3c42b5c471bf756814053e847c10f3eb003417bc523d30 LICENSE
|
| 2 |
+
a4b7f05eecdeef3f3d6182da402009c97b38907656b8f29d3ee1ae34b339e529 README.md
|
| 3 |
+
9d64f08ac126a8b37bb70c405701c4ca65baebf2904bea87ada7ca31ef8a040f data/rate_annotations/test.jsonl
|
| 4 |
+
7abc29207d16ad8114108c7f6478970d8ad38d03594e0f7370cf221bff9d3a35 prompts/README.md
|
| 5 |
+
9767fc810fba74537ea3c5283cd89062db40bb11a36e58f3b675182e9077c8c7 prompts/additional/orbit_sc_sbs.json
|
| 6 |
+
97747504de288eb78119cb26d43959c4d3467234b737fadc27b3a2ac55d13c03 prompts/additional/orbit_sc_sbs.md
|
| 7 |
+
824486e87e874501d923f19d8b2e0cbffddd0d0ef92541c08203fe0f2c0f397b prompts/additional/rate_full_with_examples_and_coverage_k2.json
|
| 8 |
+
0588a6884689f6a784f6e03809af1c47fa9596907bdd926231b0c2d0925ac83a prompts/additional/rate_full_with_examples_and_coverage_k2.md
|
| 9 |
+
d5b57f41b9a0e6290a8a2c13fc3c62f9728c9747756c7b76525757d1c82d0e02 prompts/esa_ae.json
|
| 10 |
+
fd0fef56d5f81245e8cc8d8f67706aa247b0f3b5f4ffe671f9445f8e9915ad87 prompts/esa_ae.md
|
| 11 |
+
ea208a67a4027f1949ea6c870ae395bbbc415dcfa056feab47ffd6464675ffa5 prompts/mqm_ae.json
|
| 12 |
+
9fa19a29030e737dff53b55d91f347dcc355af7951fad703f82dd0809e3bd65c prompts/mqm_ae.md
|
| 13 |
+
491d02761bf9bcf747a1f288dc8ba6d479bdf813c6578d550d3016ea4f84f601 prompts/orbit_mc_fuser.json
|
| 14 |
+
bd701419859f5a5ae95477af3eb794631b3e96b00189de8f619756d764552ea0 prompts/orbit_mc_fuser.md
|
| 15 |
+
25ce421aa5e524d32f4faaebf0a1cb39ec032286d40e5979576c514f322acb02 prompts/orbit_sc.json
|
| 16 |
+
aa08668ec2d349577c0c1b1e1601dc7be21152a7483733da543b98b6dcf31c4b prompts/orbit_sc.md
|
| 17 |
+
ef0c0f7fe0f0513353f5236cee43a94c58fb813e7534a871945f8de68d3c5cf9 prompts/prompts.jsonl
|
| 18 |
+
3f1499b21b6c5c8f8d04cad92eded3a512264a9356a77363516ec41915ba6c20 prompts/rate_brief.json
|
| 19 |
+
88458baffcec3cfe9da2379f95e0ec25e1d61f220d24a138c674d318458b3d08 prompts/rate_brief.md
|
| 20 |
+
9b164f695043518fab9a1fdef196191ccb60159a6421b8afa30dda0c6bbc0c04 prompts/rate_full.json
|
| 21 |
+
a8ddbe400e88ccc6cac41a4ccd1e07e9a865af613c8d66381aecb358ecfb86d5 prompts/rate_full.md
|
| 22 |
+
1360659471921de8ad05efee9542531267008184882ff4ae15f073dfdf623d01 prompts/rate_full_with_examples.json
|
| 23 |
+
6796d142f434f111357142101e56dd4bb979fd622f27784826b0386fa10c9b9e prompts/rate_full_with_examples.md
|
data/rate_annotations/test.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
prompts/README.md
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Evaluation prompts
|
| 2 |
+
|
| 3 |
+
This directory contains the 9 prompt bundles released with ORBIT-MT-Eval. [`prompts.jsonl`](prompts.jsonl) is the canonical machine-readable collection. The linked pages present the same ordered messages in a readable form.
|
| 4 |
+
Each linked JSON file is a minimal copy-ready `messages` object.
|
| 5 |
+
Markdown pages show prompts without separate few-shot turns as one copyable block. The JSON files preserve the exact roles and message boundaries used by the workflow.
|
| 6 |
+
|
| 7 |
+
ORBIT-SC combines the RATE brief instruction with two curated coverage demonstrations. ORBIT-MC uses that same ORBIT-SC evaluator prompt for each constituent call, then applies the released fuser prompt. RATE brief, RATE full, and RATE full + examples are zero-shot instruction-design variants.
|
| 8 |
+
|
| 9 |
+
Runtime placeholders such as `{{source}}`, `{{translation}}`, `{{translation_a}}`, `{{translation_b}}`, and `{{reports}}` are preserved.
|
| 10 |
+
|
| 11 |
+
## Main prompts
|
| 12 |
+
|
| 13 |
+
| Prompt | Purpose | JSON |
|
| 14 |
+
|---|---|---|
|
| 15 |
+
| [ORBIT-SC](orbit_sc.md) | RATE brief with curated coverage demonstrations, k=2 (ORBIT-SC) | [messages](orbit_sc.json) |
|
| 16 |
+
| [ORBIT-MC fuser](orbit_mc_fuser.md) | ORBIT-MC semantic report fuser | [messages](orbit_mc_fuser.json) |
|
| 17 |
+
| [RATE brief](rate_brief.md) | RATE brief zero-shot baseline | [messages](rate_brief.json) |
|
| 18 |
+
| [RATE full](rate_full.md) | RATE full zero-shot baseline | [messages](rate_full.json) |
|
| 19 |
+
| [RATE full + examples](rate_full_with_examples.md) | RATE full instruction with embedded examples, zero-shot | [messages](rate_full_with_examples.json) |
|
| 20 |
+
| [MQM-AE](mqm_ae.md) | MQM-AE baseline | [messages](mqm_ae.json) |
|
| 21 |
+
| [ESA-AE](esa_ae.md) | ESA-AE baseline | [messages](esa_ae.json) |
|
| 22 |
+
|
| 23 |
+
## Upstream baselines
|
| 24 |
+
|
| 25 |
+
These baselines use prompts maintained by their original authors:
|
| 26 |
+
|
| 27 |
+
- [GEMBA-MQM template](https://github.com/MicrosoftTranslator/GEMBA/blob/a7a7eff8e46998447c6cbf09d06affc8f1b99ab4/gemba/gemba_mqm_utils.py#L152-L232)
|
| 28 |
+
- [GEMBA-ESA template](https://github.com/MicrosoftTranslator/GEMBA/blob/a7a7eff8e46998447c6cbf09d06affc8f1b99ab4/gemba/gemba_esa.py#L7-L82)
|
| 29 |
+
- [`unified-mqm-boosted-v5`](https://github.com/amazon-science/span-mt-metaeval/blob/ca494e2626fd85f440ba3934404d347b281bdfcc/src/mt_evaluation/autoevals/unified/unified_mqm_boosted_v5.py), called UMB-v5 in our paper, from [*Span-Level Machine Translation Meta-Evaluation*](https://arxiv.org/abs/2603.19921) and its [official companion repository](https://github.com/amazon-science/span-mt-metaeval)
|
| 30 |
+
|
| 31 |
+
## Additional paper configurations
|
| 32 |
+
|
| 33 |
+
These variants are released for the corresponding paper analyses, but are not part of the main prompt list.
|
| 34 |
+
|
| 35 |
+
| Prompt | Paper context | JSON |
|
| 36 |
+
|---|---|---|
|
| 37 |
+
| [RATE full + examples + coverage (k=2)](additional/rate_full_with_examples_and_coverage_k2.md) | Instruction/coverage experiment. | [messages](additional/rate_full_with_examples_and_coverage_k2.json) |
|
| 38 |
+
| [ORBIT-SC (SbS)](additional/orbit_sc_sbs.md) | Appendix H and the side-by-side cost run. | [messages](additional/orbit_sc_sbs.json) |
|
prompts/additional/orbit_sc_sbs.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"messages": [
|
| 3 |
+
{
|
| 4 |
+
"role": "system",
|
| 5 |
+
"content": "# Translation Error Analysis Task — Side-by-Side Evaluation\n\n## Instructions\nYou are an expert translator. Analyze the provided source text and its TWO translations (Translation A and Translation B). Identify any errors in each translation and provide overall quality scores. Evaluate each translation INDEPENDENTLY — do not let the quality of one translation influence your assessment of the other.\n\n### Error Severity Scale (1 to 5):\n- 4-5 (Major/Critical Errors)\n- 2-3 (Minor errors)\n- 1 (Negligible or arguable errors)\n## Error Types and Categories\n\n**Semantic errors** (affect meaning):\n- mistranslation: incorrect meaning or terminology\n- undertranslation: too general, lost nuances\n- overtranslation: added unjustified specifics\n- omission: deleted information\n\n**Non-semantic errors** (don't change meaning):\n- grammar: grammar, punctuation, spelling errors\n- style: awkward or unnatural wording\n- do not translate: incorrectly translated code, URLs, etc.\n- inconsistency: inconsistent terminology within text\n- ne: incorrect proper noun translation/transcription\n- fluency: unnatural flow, confusing structure\n- other: errors not fitting above categories\n\n## Final Scores (0-100):\n- **Accuracy**: How faithfully meaning is conveyed\n- **Fluency**: How natural and readable the translation is\n\n## Output Format\nFor each error, provide:\n```\nerror_comment: [explanation in target language]\nerror_span: [problematic text from translation] For omission errors, describe what is missing in brackets, e.g., \"[missing: recent]\"\nerror_severity: [1-5]\nerror_type: [semantic/non-semantic]\nerror_categories: [category name]\n```\n\nSeparate errors with two line breaks. End each report with:\n```\n## Final scores:\nAccuracy: [0-100]\nFluency: [0-100]\n```\n\nIf no errors found for a translation, output: \"No translation mistakes found.\" then Final scores.\n\n### Output Structure:\nYou MUST produce exactly two reports, clearly separated by the following delimiters:\n\n=== Report for Translation A ===\n(list all errors for Translation A using the format above, then Final scores)\n\n=== Report for Translation B ===\n(list all errors for Translation B using the format above, then Final scores)\n\nEach report must be self-contained and end with \"## Final scores:\" followed by Accuracy and Fluency values.\n"
|
| 6 |
+
},
|
| 7 |
+
{
|
| 8 |
+
"role": "user",
|
| 9 |
+
"content": "## Text for Analysis\n### Source Text\nOn 8 September 2019, Bugatti presented a limited production high performance variant of the Chiron called the Chiron Super Sport 300+ to owners taking part in the company's 110th-anniversary tour in Europe. The variant is limited to 30 examples at a net unit price of 3.5 million euros. Changes from the standard Chiron include a more powerful engine, increased top speed, and a characteristic paint scheme featuring a lacquer-coated carbon fibre body with orange stripes, echoing the aesthetics of the company's previous top speed record-breaking cars, the Veyron Super Sport World Record Edition and the Veyron Grand Sport Vitesse World Record Car.\n### Translation A\n8 сентября 2019 года Bugatti представила ограниченную серию высокопроизводительной модификации Chiron под названием Chiron Super Sport 300+ владельцам, участвовавшим в европейском туре компании по случаю её 110-летия. Модификация выпущена тиражом в 30 экземпляров по чистой цене 3,5 миллиона евро за единицу. Отличия от стандартного Chiron включают более мощный двигатель, увеличенную максимальную скорость и характерную окраску: лакокрасочное покрытие кузова из углепластика с оранжевыми полосами, что перекликается с эстетикой предыдущих рекордных автомобилей компании — Veyron Super Sport World Record Edition и Veyron Grand Sport Vitesse World Record Car.\n### Translation B\n8 сентября 2019 года компания Bugatti представила ограниченную высокопроизводительную версию Chiron под названием Chiron Super Sport 300+ владельцам, участвовавшим в европейском туре по случаю 110-летия компании. Выпуск этой версии ограничен 30 экземплярами при цене за единицу (без налогов) 3,5 миллиона евро. Отличия от стандартного Chiron включают более мощный двигатель, увеличенную максимальную скорость и характерную схему окраски: покрытый лаком кузов из углепластика с оранжевыми полосами, отсылающую к эстетике предыдущих рекордных по максимальной скорости автомобилей компании — Veyron Super Sport World Record Edition и Veyron Grand Sport Vitesse World Record Car.\n"
|
| 10 |
+
},
|
| 11 |
+
{
|
| 12 |
+
"role": "assistant",
|
| 13 |
+
"content": "=== Report for Translation A ===\n## Error Report\nerror_comment: Нарушена лексическая сочетаемость, стоит заменить на «лимитированную серию» или обыграть «ограниченный выпуск»\nerror_span: ограниченную серию\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Необходима конкретизация: владельцы чего именно имеются в виду («автомобилей / гиперкаров Bugatti»)\nerror_span: владельцам\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Опущена информация о том, что эти автомобили побили рекорды скорости + неестественная конструкция; стоит переписать фрагмент, например, «выпущенных Bugatti автомобилей, побивших рекорды скорости»\nerror_span: рекордных\nerror_severity: 5\nerror_type: Semantic\nerror_categories: undertranslation\n\nerror_comment: Дословный перевод, стоит заменить на «более мощной» или «высокоскоростной»\nerror_span: высокопроизводительной\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Неверный контекстуальный синоним: «тираж» относится к печатной продукции\nerror_span: тиражом\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Неясная отнесённость\nerror_span: что\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Дословный перевод герундия, стоит заменить на «которые принимали / приняли участие»\nerror_span: участвовавшим\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Дословный перевод, нарушающий сочетаемость и создающий двусмысленность («включать двигатель» в значении его запуска)\nerror_span: включают\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Неверный контекстуальный синоним: «единица» создаёт ассоциации с научным текстом, стоит заменить на «каждый»\nerror_span: единицу\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Конструкция переусложнена: лучше «От стандартного Chiron модель отличается более мощным двигателем...»\nerror_span: Отличия от стандартного Chiron\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Громоздкий дословный перевод; стоит разбить на смысловые группы и переписать фрагмент\nerror_span: углепластика с оранжевыми полосами\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Нарушена лексическая и контекстуальная сочетаемость, лишние значения: «лакокрасочными» бывают работы, здесь речь о покрытии кузова\nerror_span: лакокрасочное покрытие\nerror_severity: 4\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Лишний оттенок значения, стоит заменить на «особую»\nerror_span: характерную\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\n## Final scores:\nAccuracy: 78\nFluency: 57\n\n=== Report for Translation B ===\n## Error Report\nerror_comment: Не вполне ясно, с каким словом согласуется из-за того, что «схему» и «отсылающую» сильно разнесены\nerror_span: отсылающую\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Конструкция переусложнена: лучше «От стандартного Chiron модель отличается более мощным двигателем...»\nerror_span: Отличия от стандартного Chiron\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Дословный перевод, нарушающий сочетаемость и создающий двусмысленность («включать двигатель» в значении его запуска)\nerror_span: включают\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: net unit — «чистая» цена, т.е. без учёта не только налогов, но и скидок\nerror_span: (без налогов)\nerror_severity: 4\nerror_type: Semantic\nerror_categories: undertranslation\n\nerror_comment: Дословный перевод, стоит заменить на «более мощную» или «высокоскоростную»\nerror_span: высокопроизводительную\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Нарушена лексическая сочетаемость, стоит заменить на «лимитированную серию» или обыграть «ограниченный выпуск»\nerror_span: ограниченную высокопроизводительную версию\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Неестественный синтаксис (порядок слов)\nerror_span: по максимальной скорости автомобилей компании\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Громоздкий дословный перевод; стоит разбить на смысловые группы и переписать фрагмент\nerror_span: покрытый лаком кузов из углепластика с оранжевыми полосами\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Необходима конкретизация: владельцы чего именно имеются в виду («автомобилей / гиперкаров Bugatti»)\nerror_span: владельцам\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Переставить в конце + неверный контекстуальный синоним: «единица» создаёт ассоциации с научным текстом, стоит заменить на «каждый»\nerror_span: за единицу\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Лишний оттенок значения, стоит заменить на «особую»\nerror_span: характерную\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Неверно истолкован смысл оригинала: рекорд измеряется по фактическим скоростным характеристикам, а не по заявленной максимальной скорости\nerror_span: по максимальной скорости\nerror_severity: 2\nerror_type: Semantic\nerror_categories: mistranslation\n\nerror_comment: Дословный перевод герундия, стоит заменить на «которые принимали / приняли участие»\nerror_span: участвовавшим\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Неестественная конструкция; стоит переписать фрагмент, например, «выпущенных Bugatti автомобилей, побивших рекорды скорости»\nerror_span: рекордных\nerror_severity: 3\nerror_type: Semantic\nerror_categories: undertranslation\n\nerror_comment: Нарушена лексическая сочетаемость + громоздкая конструкция: «(Будет) выпущено всего...»\nerror_span: Выпуск этой версии ограничен\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\n## Final scores:\nAccuracy: 75\nFluency: 58\n"
|
| 14 |
+
},
|
| 15 |
+
{
|
| 16 |
+
"role": "user",
|
| 17 |
+
"content": "## Text for Analysis\n### Source Text\nYou, you the world! You're full of surprises, aren't you? I remember a few years back, watching some of those Resident Evil movies with Mila Jovovich and thinking, I can't get any worse than this. Yeah, it feels like a simpler time. But with 2021's final gasping breath that came out with Resident Evil: Welcome to Raccoon City. AKA, Welcome to Hell.\n### Translation A\nТы, весь мир! Ты полон сюрпризов, не так ли? Помню, несколько лет назад я смотрел некоторые фильмы из серии «Обитель зла» с Милой Йовович и думал: «Хуже уже быть не может». Да, тогда казалось, что времена были проще. Но на последнем издыхании 2021 года вышел фильм «Обитель зла: Добро пожаловать в Раккун-Сити». Он же — «Добро пожаловать в ад».\n### Translation B\nТы, ты, мир! Ты полон сюрпризов, не так ли? Помню, как несколько лет назад я смотрел фильмы из серии «Обитель зла» с Милой Йовович и думал: хуже уже быть не может. Да, кажется, что тогда все было проще. Но в 2021 году вышел фильм «Обитель зла: Раккун-Сити», и это был последний вздох. Или, другими словами, добро пожаловать в ад.\n"
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"role": "assistant",
|
| 21 |
+
"content": "=== Report for Translation A ===\n## Error Report\nerror_comment: Верное написание: Милла.\nerror_span: Милой\nerror_severity: 4\nerror_type: Semantic\nerror_categories: NE\n\nerror_comment: Близко «несколько/некоторые»\nerror_span: некоторые\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Понятно по смыслу, но калька.\nerror_span: Ты полон сюрпризов\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: В исходнике нет такого значения.\nerror_span: весь\nerror_severity: 4\nerror_type: Semantic\nerror_categories: overtranslation\n\nerror_comment: Автору кажется это сейчас, а не казалось раньше.\nerror_span: тогда казалось\nerror_severity: 3\nerror_type: Semantic\nerror_categories: mistranslation\n\nerror_comment: Теряется субъективность и ирония – то, что автор сам дает название этому фильму.\nerror_span: Он же\nerror_severity: 2\nerror_type: Semantic\nerror_categories: undertranslation\n\nerror_comment: Слишком буквальный перевод, плохо считывается как обращение к миру.\nerror_span: Ты, весь мир\nerror_severity: 5\nerror_type: Semantic\nerror_categories: mistranslation\n\nerror_comment: Tag question, не несёт смысла, можно не переводить. Звучит неестественно.\nerror_span: не так ли\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Речь не о том, что времена были проще, а о том, что они были лучше.\nerror_span: времена были проще\nerror_severity: 3\nerror_type: Semantic\nerror_categories: mistranslation\n\nerror_comment: Верное название фильма в российском прокате: «Обитель зла: Раккун-Сити»\nerror_span: «Обитель зла: Добро пожаловать в Раккун-Сити»\nerror_severity: 4\nerror_type: Semantic\nerror_categories: NE\n\nerror_comment: Созвучные слова, неудобно для произношения.\nerror_span: «Хуже уже\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\n## Final scores:\nAccuracy: 60\nFluency: 77\n\n=== Report for Translation B ===\n## Error Report\nerror_comment: В переводе нет информации о том, что автор посмотрел несколько фильмов.\nerror_span: some\nerror_severity: 2\nerror_type: Semantic\nerror_categories: omission\n\nerror_comment: Понятно по смыслу, но калька.\nerror_span: Ты полон сюрпризов\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Название фильма в русскоязычном прокате передано правильно (просто «Обитель зла: Раккун-Сити»), но из-за этого здесь потерялась игра слов с «Добро пожаловать».\nerror_span: добро пожаловать в ад\nerror_severity: 2\nerror_type: Semantic\nerror_categories: undertranslation\n\nerror_comment: Дословный перевод, звучит неестественно.\nerror_span: Ты, ты, мир\nerror_severity: 4\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Созвучные слова, неудобно для произношения.\nerror_span: хуже уже\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Фразу нужно поставить раньше, иначе теряется связка со следующим предложением.\nerror_span: и это был последний вздох\nerror_severity: 4\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Последний вздох четко приравнивается к фильму, хотя в исходнике нет такой однозначности.\nerror_span: это был\nerror_severity: 4\nerror_type: Semantic\nerror_categories: mistranslation\n\nerror_comment: Верное написание: Милла.\nerror_span: Милой\nerror_severity: 4\nerror_type: Semantic\nerror_categories: NE\n\nerror_comment: Речь не о том, что времена были проще, а о том, что они были лучше.\nerror_span: тогда все было проще\nerror_severity: 3\nerror_type: Semantic\nerror_categories: mistranslation\n\nerror_comment: Tag question, не несёт смысла, можно не переводить. Звучит неестественно.\nerror_span: не так ли\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\n## Final scores:\nAccuracy: 57\nFluency: 74\n"
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"role": "user",
|
| 25 |
+
"content": "### Source Text\n{{source}}\n\n### Translation A\n{{translation_a}}\n\n### Translation B\n{{translation_b}}\n"
|
| 26 |
+
}
|
| 27 |
+
]
|
| 28 |
+
}
|
prompts/additional/orbit_sc_sbs.md
ADDED
|
@@ -0,0 +1,428 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# ORBIT-SC (SbS)
|
| 2 |
+
|
| 3 |
+
## 1. `system`
|
| 4 |
+
|
| 5 |
+
````text
|
| 6 |
+
# Translation Error Analysis Task — Side-by-Side Evaluation
|
| 7 |
+
|
| 8 |
+
## Instructions
|
| 9 |
+
You are an expert translator. Analyze the provided source text and its TWO translations (Translation A and Translation B). Identify any errors in each translation and provide overall quality scores. Evaluate each translation INDEPENDENTLY — do not let the quality of one translation influence your assessment of the other.
|
| 10 |
+
|
| 11 |
+
### Error Severity Scale (1 to 5):
|
| 12 |
+
- 4-5 (Major/Critical Errors)
|
| 13 |
+
- 2-3 (Minor errors)
|
| 14 |
+
- 1 (Negligible or arguable errors)
|
| 15 |
+
## Error Types and Categories
|
| 16 |
+
|
| 17 |
+
**Semantic errors** (affect meaning):
|
| 18 |
+
- mistranslation: incorrect meaning or terminology
|
| 19 |
+
- undertranslation: too general, lost nuances
|
| 20 |
+
- overtranslation: added unjustified specifics
|
| 21 |
+
- omission: deleted information
|
| 22 |
+
|
| 23 |
+
**Non-semantic errors** (don't change meaning):
|
| 24 |
+
- grammar: grammar, punctuation, spelling errors
|
| 25 |
+
- style: awkward or unnatural wording
|
| 26 |
+
- do not translate: incorrectly translated code, URLs, etc.
|
| 27 |
+
- inconsistency: inconsistent terminology within text
|
| 28 |
+
- ne: incorrect proper noun translation/transcription
|
| 29 |
+
- fluency: unnatural flow, confusing structure
|
| 30 |
+
- other: errors not fitting above categories
|
| 31 |
+
|
| 32 |
+
## Final Scores (0-100):
|
| 33 |
+
- **Accuracy**: How faithfully meaning is conveyed
|
| 34 |
+
- **Fluency**: How natural and readable the translation is
|
| 35 |
+
|
| 36 |
+
## Output Format
|
| 37 |
+
For each error, provide:
|
| 38 |
+
```
|
| 39 |
+
error_comment: [explanation in target language]
|
| 40 |
+
error_span: [problematic text from translation] For omission errors, describe what is missing in brackets, e.g., "[missing: recent]"
|
| 41 |
+
error_severity: [1-5]
|
| 42 |
+
error_type: [semantic/non-semantic]
|
| 43 |
+
error_categories: [category name]
|
| 44 |
+
```
|
| 45 |
+
|
| 46 |
+
Separate errors with two line breaks. End each report with:
|
| 47 |
+
```
|
| 48 |
+
## Final scores:
|
| 49 |
+
Accuracy: [0-100]
|
| 50 |
+
Fluency: [0-100]
|
| 51 |
+
```
|
| 52 |
+
|
| 53 |
+
If no errors found for a translation, output: "No translation mistakes found." then Final scores.
|
| 54 |
+
|
| 55 |
+
### Output Structure:
|
| 56 |
+
You MUST produce exactly two reports, clearly separated by the following delimiters:
|
| 57 |
+
|
| 58 |
+
=== Report for Translation A ===
|
| 59 |
+
(list all errors for Translation A using the format above, then Final scores)
|
| 60 |
+
|
| 61 |
+
=== Report for Translation B ===
|
| 62 |
+
(list all errors for Translation B using the format above, then Final scores)
|
| 63 |
+
|
| 64 |
+
Each report must be self-contained and end with "## Final scores:" followed by Accuracy and Fluency values.
|
| 65 |
+
````
|
| 66 |
+
|
| 67 |
+
## 2. `user`
|
| 68 |
+
|
| 69 |
+
```text
|
| 70 |
+
## Text for Analysis
|
| 71 |
+
### Source Text
|
| 72 |
+
On 8 September 2019, Bugatti presented a limited production high performance variant of the Chiron called the Chiron Super Sport 300+ to owners taking part in the company's 110th-anniversary tour in Europe. The variant is limited to 30 examples at a net unit price of 3.5 million euros. Changes from the standard Chiron include a more powerful engine, increased top speed, and a characteristic paint scheme featuring a lacquer-coated carbon fibre body with orange stripes, echoing the aesthetics of the company's previous top speed record-breaking cars, the Veyron Super Sport World Record Edition and the Veyron Grand Sport Vitesse World Record Car.
|
| 73 |
+
### Translation A
|
| 74 |
+
8 сентября 2019 года Bugatti представила ограниченную серию высокопроизводительной модификации Chiron под названием Chiron Super Sport 300+ владельцам, участвовавшим в европейском туре компании по случаю её 110-летия. Модификация выпущена тиражом в 30 экземпляров по чистой цене 3,5 миллиона евро за единицу. Отличия от стандартного Chiron включают более мощный двигатель, увеличенную максимальную скорость и характерную окраску: лакокрасочное покрытие кузова из углепластика с оранжевыми полосами, что перекликается с эстетикой предыдущих рекордных автомобилей компании — Veyron Super Sport World Record Edition и Veyron Grand Sport Vitesse World Record Car.
|
| 75 |
+
### Translation B
|
| 76 |
+
8 сентября 2019 года компания Bugatti представила ограниченную высокопроизводительную версию Chiron под названием Chiron Super Sport 300+ владельцам, участвовавшим в европейском туре по случаю 110-летия компании. Выпуск этой версии ограничен 30 экземплярами при цене за единицу (без налогов) 3,5 миллиона евро. Отличия от стандартного Chiron включают более мощный двигатель, увеличенную максимальную скорость и характерную схему окраски: покрытый лаком кузов из углепластика с оранжевыми полосами, отсылающую к эстетике предыдущих рекордных по максимальной скорости автомобилей компании — Veyron Super Sport World Record Edition и Veyron Grand Sport Vitesse World Record Car.
|
| 77 |
+
```
|
| 78 |
+
|
| 79 |
+
## 3. `assistant`
|
| 80 |
+
|
| 81 |
+
```text
|
| 82 |
+
=== Report for Translation A ===
|
| 83 |
+
## Error Report
|
| 84 |
+
error_comment: Нарушена лексическая сочетаемость, стоит заменить на «лимитированную серию» или обыграть «ограниченный выпуск»
|
| 85 |
+
error_span: ограниченную серию
|
| 86 |
+
error_severity: 3
|
| 87 |
+
error_type: Non-semantic
|
| 88 |
+
error_categories: fluency
|
| 89 |
+
|
| 90 |
+
error_comment: Необходима конкретизация: владельцы чего именно имеются в виду («автомобилей / гиперкаров Bugatti»)
|
| 91 |
+
error_span: владельцам
|
| 92 |
+
error_severity: 3
|
| 93 |
+
error_type: Non-semantic
|
| 94 |
+
error_categories: fluency
|
| 95 |
+
|
| 96 |
+
error_comment: Опущена информация о том, что эти автомобили побили рекорды скорости + неестественная конструкция; стоит переписать фрагмент, например, «выпущенных Bugatti автомобилей, побивших рекорды скорости»
|
| 97 |
+
error_span: рекордных
|
| 98 |
+
error_severity: 5
|
| 99 |
+
error_type: Semantic
|
| 100 |
+
error_categories: undertranslation
|
| 101 |
+
|
| 102 |
+
error_comment: Дословный перевод, стоит заменить на «более мощной» или «высокоскоростной»
|
| 103 |
+
error_span: высокопроизводительной
|
| 104 |
+
error_severity: 3
|
| 105 |
+
error_type: Non-semantic
|
| 106 |
+
error_categories: fluency
|
| 107 |
+
|
| 108 |
+
error_comment: Неверный контекстуальный синоним: «тираж» относится к печатной продукции
|
| 109 |
+
error_span: тиражом
|
| 110 |
+
error_severity: 3
|
| 111 |
+
error_type: Non-semantic
|
| 112 |
+
error_categories: fluency
|
| 113 |
+
|
| 114 |
+
error_comment: Неясная отнесённость
|
| 115 |
+
error_span: что
|
| 116 |
+
error_severity: 2
|
| 117 |
+
error_type: Non-semantic
|
| 118 |
+
error_categories: fluency
|
| 119 |
+
|
| 120 |
+
error_comment: Дословный перевод герундия, стоит заменить на «которые принимали / приняли участие»
|
| 121 |
+
error_span: участвовавшим
|
| 122 |
+
error_severity: 2
|
| 123 |
+
error_type: Non-semantic
|
| 124 |
+
error_categories: fluency
|
| 125 |
+
|
| 126 |
+
error_comment: Дословный перевод, нарушающий сочетаемость и создающий двусмысленность («включать двигатель» в значении его запуска)
|
| 127 |
+
error_span: включают
|
| 128 |
+
error_severity: 2
|
| 129 |
+
error_type: Non-semantic
|
| 130 |
+
error_categories: fluency
|
| 131 |
+
|
| 132 |
+
error_comment: Неверный контекстуальный синоним: «единица» создаёт ассоциации с научным текстом, стоит заменить на «каждый»
|
| 133 |
+
error_span: единицу
|
| 134 |
+
error_severity: 3
|
| 135 |
+
error_type: Non-semantic
|
| 136 |
+
error_categories: fluency
|
| 137 |
+
|
| 138 |
+
error_comment: Конструкция переусложнена: лучше «От стандартного Chiron модель отличается более мощным двигателем...»
|
| 139 |
+
error_span: Отличия от стандартного Chiron
|
| 140 |
+
error_severity: 2
|
| 141 |
+
error_type: Non-semantic
|
| 142 |
+
error_categories: fluency
|
| 143 |
+
|
| 144 |
+
error_comment: Громоздкий дословный перевод; стоит разбить на смысловые группы и переписать фрагмент
|
| 145 |
+
error_span: углепластика с оранжевыми полосами
|
| 146 |
+
error_severity: 2
|
| 147 |
+
error_type: Non-semantic
|
| 148 |
+
error_categories: fluency
|
| 149 |
+
|
| 150 |
+
error_comment: Нарушена лексическая и контекстуальная сочетаемость, лишние значения: «лакокрасочными» бывают работы, здесь речь о покрытии кузова
|
| 151 |
+
error_span: лакокрасочное покрытие
|
| 152 |
+
error_severity: 4
|
| 153 |
+
error_type: Non-semantic
|
| 154 |
+
error_categories: fluency
|
| 155 |
+
|
| 156 |
+
error_comment: Лишний оттенок значения, стоит заменить на «особую»
|
| 157 |
+
error_span: характерную
|
| 158 |
+
error_severity: 2
|
| 159 |
+
error_type: Non-semantic
|
| 160 |
+
error_categories: fluency
|
| 161 |
+
|
| 162 |
+
## Final scores:
|
| 163 |
+
Accuracy: 78
|
| 164 |
+
Fluency: 57
|
| 165 |
+
|
| 166 |
+
=== Report for Translation B ===
|
| 167 |
+
## Error Report
|
| 168 |
+
error_comment: Не вполне ясно, с каким словом согласуется из-за того, что «схему» и «отсылающую» сильно разнесены
|
| 169 |
+
error_span: отсылающую
|
| 170 |
+
error_severity: 2
|
| 171 |
+
error_type: Non-semantic
|
| 172 |
+
error_categories: fluency
|
| 173 |
+
|
| 174 |
+
error_comment: Конструкция переусложнена: лучше «От стандартного Chiron модель отличается более мощным двигателем...»
|
| 175 |
+
error_span: Отличия от стандартного Chiron
|
| 176 |
+
error_severity: 2
|
| 177 |
+
error_type: Non-semantic
|
| 178 |
+
error_categories: fluency
|
| 179 |
+
|
| 180 |
+
error_comment: Дословный перевод, нарушающий сочетаемость и создающий двусмысленность («включать двигатель» в значении его запуска)
|
| 181 |
+
error_span: включают
|
| 182 |
+
error_severity: 2
|
| 183 |
+
error_type: Non-semantic
|
| 184 |
+
error_categories: fluency
|
| 185 |
+
|
| 186 |
+
error_comment: net unit — «чистая» цена, т.е. без учёта не только налогов, но и скидок
|
| 187 |
+
error_span: (без налогов)
|
| 188 |
+
error_severity: 4
|
| 189 |
+
error_type: Semantic
|
| 190 |
+
error_categories: undertranslation
|
| 191 |
+
|
| 192 |
+
error_comment: Дословный перевод, стоит заменить на «более мощную» или «высокоскоростную»
|
| 193 |
+
error_span: высокопроизводительную
|
| 194 |
+
error_severity: 3
|
| 195 |
+
error_type: Non-semantic
|
| 196 |
+
error_categories: fluency
|
| 197 |
+
|
| 198 |
+
error_comment: Нарушена лексическая сочетаемость, стоит заменить на «лимитированную серию» или обыграть «ограниченный выпуск»
|
| 199 |
+
error_span: ограниченную высокопроизводительную версию
|
| 200 |
+
error_severity: 2
|
| 201 |
+
error_type: Non-semantic
|
| 202 |
+
error_categories: fluency
|
| 203 |
+
|
| 204 |
+
error_comment: Неестественный синтаксис (порядок слов)
|
| 205 |
+
error_span: по максимальной скорости автомобилей компании
|
| 206 |
+
error_severity: 2
|
| 207 |
+
error_type: Non-semantic
|
| 208 |
+
error_categories: fluency
|
| 209 |
+
|
| 210 |
+
error_comment: Громоздкий дословный перевод; стоит разбить на смысловые группы и переписать фрагмент
|
| 211 |
+
error_span: покрытый лаком кузов из углепластика с оранжевыми полосами
|
| 212 |
+
error_severity: 2
|
| 213 |
+
error_type: Non-semantic
|
| 214 |
+
error_categories: fluency
|
| 215 |
+
|
| 216 |
+
error_comment: Необходима конкретизация: владельцы чего именно имеются в виду («автомобилей / гиперкаров Bugatti»)
|
| 217 |
+
error_span: владельцам
|
| 218 |
+
error_severity: 3
|
| 219 |
+
error_type: Non-semantic
|
| 220 |
+
error_categories: fluency
|
| 221 |
+
|
| 222 |
+
error_comment: Переставить в конце + неверный контекстуальный синоним: «единица» создаёт ассоциации с научным текстом, стоит заменить на «каждый»
|
| 223 |
+
error_span: за единицу
|
| 224 |
+
error_severity: 3
|
| 225 |
+
error_type: Non-semantic
|
| 226 |
+
error_categories: fluency
|
| 227 |
+
|
| 228 |
+
error_comment: Лишний оттенок значения, стоит заменить на «особую»
|
| 229 |
+
error_span: характерную
|
| 230 |
+
error_severity: 2
|
| 231 |
+
error_type: Non-semantic
|
| 232 |
+
error_categories: fluency
|
| 233 |
+
|
| 234 |
+
error_comment: Неверно истолкован смысл оригинала: рекорд измеряется по фактическим скоростным характеристикам, а не по заявленной максимальной скорости
|
| 235 |
+
error_span: по максимальной скорости
|
| 236 |
+
error_severity: 2
|
| 237 |
+
error_type: Semantic
|
| 238 |
+
error_categories: mistranslation
|
| 239 |
+
|
| 240 |
+
error_comment: Дословный перевод герундия, стоит заменить на «которые принимали / приняли участие»
|
| 241 |
+
error_span: участвовавшим
|
| 242 |
+
error_severity: 2
|
| 243 |
+
error_type: Non-semantic
|
| 244 |
+
error_categories: fluency
|
| 245 |
+
|
| 246 |
+
error_comment: Неестественная конструкция; стоит переписать фрагмент, например, «выпущенных Bugatti автомобилей, побивших рекорды скорости»
|
| 247 |
+
error_span: рекордных
|
| 248 |
+
error_severity: 3
|
| 249 |
+
error_type: Semantic
|
| 250 |
+
error_categories: undertranslation
|
| 251 |
+
|
| 252 |
+
error_comment: Нарушена лексическая сочетаемость + громоздкая конструкция: «(Будет) выпущено всего...»
|
| 253 |
+
error_span: Выпуск этой версии ограничен
|
| 254 |
+
error_severity: 2
|
| 255 |
+
error_type: Non-semantic
|
| 256 |
+
error_categories: fluency
|
| 257 |
+
|
| 258 |
+
## Final scores:
|
| 259 |
+
Accuracy: 75
|
| 260 |
+
Fluency: 58
|
| 261 |
+
```
|
| 262 |
+
|
| 263 |
+
## 4. `user`
|
| 264 |
+
|
| 265 |
+
```text
|
| 266 |
+
## Text for Analysis
|
| 267 |
+
### Source Text
|
| 268 |
+
You, you the world! You're full of surprises, aren't you? I remember a few years back, watching some of those Resident Evil movies with Mila Jovovich and thinking, I can't get any worse than this. Yeah, it feels like a simpler time. But with 2021's final gasping breath that came out with Resident Evil: Welcome to Raccoon City. AKA, Welcome to Hell.
|
| 269 |
+
### Translation A
|
| 270 |
+
Ты, весь мир! Ты полон сюрпризов, не так ли? Помню, несколько лет назад я смотрел некоторые фильмы из серии «Обитель зла» с Милой Йовович и думал: «Хуже уже быть не может». Да, тогда казалось, что времена были проще. Но на последн��м издыхании 2021 года вышел фильм «Обитель зла: Добро пожаловать в Раккун-Сити». Он же — «Добро пожаловать в ад».
|
| 271 |
+
### Translation B
|
| 272 |
+
Ты, ты, мир! Ты полон сюрпризов, не так ли? Помню, как несколько лет назад я смотрел фильмы из серии «Обитель зла» с Милой Йовович и думал: хуже уже быть не может. Да, кажется, что тогда все было проще. Но в 2021 году вышел фильм «Обитель зла: Раккун-Сити», и это был последний вздох. Или, другими словами, добро пожаловать в ад.
|
| 273 |
+
```
|
| 274 |
+
|
| 275 |
+
## 5. `assistant`
|
| 276 |
+
|
| 277 |
+
```text
|
| 278 |
+
=== Report for Translation A ===
|
| 279 |
+
## Error Report
|
| 280 |
+
error_comment: Верное написание: Милла.
|
| 281 |
+
error_span: Милой
|
| 282 |
+
error_severity: 4
|
| 283 |
+
error_type: Semantic
|
| 284 |
+
error_categories: NE
|
| 285 |
+
|
| 286 |
+
error_comment: Близко «несколько/некоторые»
|
| 287 |
+
error_span: некоторые
|
| 288 |
+
error_severity: 2
|
| 289 |
+
error_type: Non-semantic
|
| 290 |
+
error_categories: fluency
|
| 291 |
+
|
| 292 |
+
error_comment: Понятно по смыслу, но калька.
|
| 293 |
+
error_span: Ты полон сюрпризов
|
| 294 |
+
error_severity: 2
|
| 295 |
+
error_type: Non-semantic
|
| 296 |
+
error_categories: fluency
|
| 297 |
+
|
| 298 |
+
error_comment: В исходнике нет такого значения.
|
| 299 |
+
error_span: весь
|
| 300 |
+
error_severity: 4
|
| 301 |
+
error_type: Semantic
|
| 302 |
+
error_categories: overtranslation
|
| 303 |
+
|
| 304 |
+
error_comment: Автору кажется это сейчас, а не казалось раньше.
|
| 305 |
+
error_span: тогда казалось
|
| 306 |
+
error_severity: 3
|
| 307 |
+
error_type: Semantic
|
| 308 |
+
error_categories: mistranslation
|
| 309 |
+
|
| 310 |
+
error_comment: Теряется субъективность и ирония – то, что автор сам дает название этому фильму.
|
| 311 |
+
error_span: Он же
|
| 312 |
+
error_severity: 2
|
| 313 |
+
error_type: Semantic
|
| 314 |
+
error_categories: undertranslation
|
| 315 |
+
|
| 316 |
+
error_comment: Слишком буквальный перевод, плохо считывается как обращение к миру.
|
| 317 |
+
error_span: Ты, весь мир
|
| 318 |
+
error_severity: 5
|
| 319 |
+
error_type: Semantic
|
| 320 |
+
error_categories: mistranslation
|
| 321 |
+
|
| 322 |
+
error_comment: Tag question, не несёт смысла, можно не переводить. Звучит неестественно.
|
| 323 |
+
error_span: не так ли
|
| 324 |
+
error_severity: 3
|
| 325 |
+
error_type: Non-semantic
|
| 326 |
+
error_categories: fluency
|
| 327 |
+
|
| 328 |
+
error_comment: Речь не о том, что времена были проще, а о том, что они были лучше.
|
| 329 |
+
error_span: времена были проще
|
| 330 |
+
error_severity: 3
|
| 331 |
+
error_type: Semantic
|
| 332 |
+
error_categories: mistranslation
|
| 333 |
+
|
| 334 |
+
error_comment: Верное название фильма в российском прокате: «Обитель зла: Раккун-Сити»
|
| 335 |
+
error_span: «Обитель зла: Добро пожаловать в Раккун-Сити»
|
| 336 |
+
error_severity: 4
|
| 337 |
+
error_type: Semantic
|
| 338 |
+
error_categories: NE
|
| 339 |
+
|
| 340 |
+
error_comment: Созвучные слова, неудобно для произношения.
|
| 341 |
+
error_span: «Хуже уже
|
| 342 |
+
error_severity: 2
|
| 343 |
+
error_type: Non-semantic
|
| 344 |
+
error_categories: fluency
|
| 345 |
+
|
| 346 |
+
## Final scores:
|
| 347 |
+
Accuracy: 60
|
| 348 |
+
Fluency: 77
|
| 349 |
+
|
| 350 |
+
=== Report for Translation B ===
|
| 351 |
+
## Error Report
|
| 352 |
+
error_comment: В переводе нет информации о том, что автор посмотрел несколько фильмов.
|
| 353 |
+
error_span: some
|
| 354 |
+
error_severity: 2
|
| 355 |
+
error_type: Semantic
|
| 356 |
+
error_categories: omission
|
| 357 |
+
|
| 358 |
+
error_comment: Понятно по смыслу, но калька.
|
| 359 |
+
error_span: Ты полон сюрпризов
|
| 360 |
+
error_severity: 2
|
| 361 |
+
error_type: Non-semantic
|
| 362 |
+
error_categories: fluency
|
| 363 |
+
|
| 364 |
+
error_comment: Название фильма в русскоязычном прокате передано правильно (просто «Обитель зла: Раккун-Сити»), но из-за этого здесь потерялась игра слов с «Добро пожаловать».
|
| 365 |
+
error_span: добро пожаловать в ад
|
| 366 |
+
error_severity: 2
|
| 367 |
+
error_type: Semantic
|
| 368 |
+
error_categories: undertranslation
|
| 369 |
+
|
| 370 |
+
error_comment: Дословный перевод, звучит неестественно.
|
| 371 |
+
error_span: Ты, ты, мир
|
| 372 |
+
error_severity: 4
|
| 373 |
+
error_type: Non-semantic
|
| 374 |
+
error_categories: fluency
|
| 375 |
+
|
| 376 |
+
error_comment: Созвучные слова, неудобно для произношения.
|
| 377 |
+
error_span: хуже уже
|
| 378 |
+
error_severity: 2
|
| 379 |
+
error_type: Non-semantic
|
| 380 |
+
error_categories: fluency
|
| 381 |
+
|
| 382 |
+
error_comment: Фразу нужно поставить раньше, иначе теряется связка со следующим предложением.
|
| 383 |
+
error_span: и это был последний вздох
|
| 384 |
+
error_severity: 4
|
| 385 |
+
error_type: Non-semantic
|
| 386 |
+
error_categories: fluency
|
| 387 |
+
|
| 388 |
+
error_comment: Последний вздох четко приравнивается к фильму, хотя в исходнике нет такой однозначности.
|
| 389 |
+
error_span: это был
|
| 390 |
+
error_severity: 4
|
| 391 |
+
error_type: Semantic
|
| 392 |
+
error_categories: mistranslation
|
| 393 |
+
|
| 394 |
+
error_comment: Верное написание: Милла.
|
| 395 |
+
error_span: Милой
|
| 396 |
+
error_severity: 4
|
| 397 |
+
error_type: Semantic
|
| 398 |
+
error_categories: NE
|
| 399 |
+
|
| 400 |
+
error_comment: Речь не о том, что времена были проще, а о том, что они были лучше.
|
| 401 |
+
error_span: тогда все было проще
|
| 402 |
+
error_severity: 3
|
| 403 |
+
error_type: Semantic
|
| 404 |
+
error_categories: mistranslation
|
| 405 |
+
|
| 406 |
+
error_comment: Tag question, не несёт смысла, можно не переводить. Звучит неестественно.
|
| 407 |
+
error_span: не так ли
|
| 408 |
+
error_severity: 3
|
| 409 |
+
error_type: Non-semantic
|
| 410 |
+
error_categories: fluency
|
| 411 |
+
|
| 412 |
+
## Final scores:
|
| 413 |
+
Accuracy: 57
|
| 414 |
+
Fluency: 74
|
| 415 |
+
```
|
| 416 |
+
|
| 417 |
+
## 6. `user`
|
| 418 |
+
|
| 419 |
+
```text
|
| 420 |
+
### Source Text
|
| 421 |
+
{{source}}
|
| 422 |
+
|
| 423 |
+
### Translation A
|
| 424 |
+
{{translation_a}}
|
| 425 |
+
|
| 426 |
+
### Translation B
|
| 427 |
+
{{translation_b}}
|
| 428 |
+
```
|
prompts/additional/rate_full_with_examples_and_coverage_k2.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"messages": [
|
| 3 |
+
{
|
| 4 |
+
"role": "system",
|
| 5 |
+
"content": "# Translation Error Analysis Task\n\n## Instructions\nYou are an expert translator and linguistic reviewer. Your task is to carefully analyze the provided original text and its translation. You must identify and clearly explain any errors or issues according to the guidelines below. You will be provided with the source, translation, and previous context source(optional) and translation. You must evaluate only the given translation, but you should read the context carefully to identify any mistakes related to it.\n\n## Guidelines\n\n### Error Severity Scale (1 to 5):\n- 4-5 (Major/Critical Errors): These errors significantly distort meaning, lead readers to misunderstand the original message, or are highly inappropriate for the target audience.\nExamples: Incorrect key terminology, significant mistranslation, adding incorrect or misleading information, serious grammatical errors that change meaning, cultural insensitivity/offensiveness, translation that is highly misleading or severely disrupts understanding.\n\n- 2-3 (Minor errors): These errors negatively impact readability, fluency, style, or grammar without seriously distorting essential information.\nExamples: Awkward phrasing, unnatural word choices, literal translations, incorrect but minor grammar/punctuation mistakes, inconsistent style or terminology that slightly impacts readability without compromising the main message.\n\n- 1 (Negligible or arguable errors): Issues that some translators may interpret differently, debatable stylistic nuances, or very minor mistakes that do not meaningfully affect text quality.\nExample: Using a suitable but less preferable word choice or style variation that is still acceptable in general translation practice.\n\n### Types and Categories of Translation Errors\nAll errors must be classified into one of two main types: Semantic or Non-semantic. Within each type, select the appropriate specific category.\n\n#### Semantic Errors (errors affecting meaning):\n- Mistranslation: Incorrect translation of meaning. This category covers:\n - Significantly incorrect meaning or terminology.\n - Ambiguity that leads to confusion or misunderstanding.\n - Literal translation that misses the intended meaning of idioms/phrases.\n - Errors in concepts critical to understanding the content.\nExamples:\nSource: \"Now we have three wires: an actual input, its inverted counterpart and sign of the input.\"\nTranslation: \"Теперь у нас есть три провода: основной вход, его инвертированная копия и сигнал на входе.\"\nProblems here: \"основной\" (\"main\") is incorrect for \"actual\"; \"копия\" (\"copy\") is incorrect for \"counterpart\"; \"сигнал на входе\" (\"input signal\") is a mistranslation for \"sign of the input.\"\n\nSource: \"...digital rights management...\"\nTranslation: \"управление цифровыми правами\" is actually \"digital rights management,\" but the better known term in Russian is \"защита цифровых прав\" or \"защита авторских прав на цифровой контент.\"\n\nSource: \"...We've all been there.\"\nTranslation: \"Мы все были там.\" (\"там\" literally means “there” physically, while proper idiomatic meaning would be \"Мы все через это проходили.\")\n\n- Undertranslation: Too general translation; details and important nuances of the original message are lost or diluted significantly.\nExample:\nSource: \"Drain the fluid from the FRONT differential.\"\nTranslation: \"Слейте жидкость...\"\nProblem: Missing critical detail (\"differential fluid\" not specified).\n\n- Overtranslation: Unjustified or overly-specific translation of a general term. Adding unnecessary specific information that was not in the source.\nExample: Translating general \"meal\" specifically as \"обед\" (\"lunch\").\n\n- Omission: Deleting words, phrases, or information from the original text without a justified reason. Sometimes, minor omissions with very limited impact may be negligible or minor severity.\nExample:\nSource: \"...you need much more than looking up typical interview questions...\"\nTranslation: \"...понадобится больше, чем просто погуглить стандартные вопросы.\"\nProblem: \"much more\" should specify \"значительно больше\" or \"заметно больше.\"\n\n#### Non-Semantic Errors (errors that don't directly change meaning):\n- Grammar: Grammar and punctuation errors, spelling mistakes, incorrect capitalization, and typos that don't dramatically affect meaning but lower translation quality and professionalism.\n\n- Style: Stylistically awkward or unnatural wording choices. Usually minor but can be critical if style is severely affected.\nExample: Translating \"apparel company\" as \"компания по изготовлению готового платья\" (this is stylistically unnatural).\n\n- Do Not Translate: Items or information that should remain untranslated (e.g., code snippets, URLs, bibliographic references, names of buttons in software). Undue translation of such elements is considered an error.\nExample: Translating code snippets incorrectly.\n\n- Inconsistency: Variations or inconsistent translations within the same text (terminology, names, gender forms). Usually minor but can become major if it causes confusion.\nExample: Translating \"squash\" inconsistently as both \"тыква\" and \"кабачок\" within the same context.\n\n- Named Entity (NE): Incorrect translation or transcription of proper nouns (person names, place names, brand/company names). Usually minor in European names if the overall identity is recognizable, major if the name becomes unrecognizable or for languages with complex transliteration rules (e.g., Asian languages). If a standard known translation exists and is ignored, consider severity 4.\n\n- Fluency: Issues affecting readability, smoothness, or natural flow of sentences, confusing structure/order of words. This category includes unnatural sentence constructions or very awkward phrasing.\nExample:\nSource: \"Rachel Brosnahan to play Lois Lane alongside David Corenswet's Clark Kent.\"\nTranslation: \"Рэйчел Броснахэн сыграет Лоис Лэйн вместе с Кларком Кентом в исполнении Дэвида Коренсвета...\"\nProblem: This implies Brosnahan will play both Lois Lane and Clark Kent, causing misunderstanding of sentence structure meanings.\n\n#### Other\n- Other: Errors that clearly impact the quality or correctness of the translation but do not logically fit into any of the previously defined semantic or non-semantic categories. This category should be used sparingly—only when no other defined categories apply. When using \"other,\" you should explain in detail why the existing categories do not fit, clearly describing the nature and impact of the identified issue.\n\n### Final Accuracy and Fluency Assessment\nAfter identifying all errors, you must provide overall scores for accuracy and fluency on a scale from 0 to 100:\n\nAccuracy (0-100): Measures how faithfully the translation conveys the meaning of the source text. Consider whether all information is correctly transferred, whether key concepts are accurately represented, and whether there are any semantic distortions.\n90-100: Near perfect accuracy with minimal or no semantic errors\n70-89: Good accuracy with few minor meaning issues\n50-69: Moderate accuracy with some notable meaning issues\n30-49: Poor accuracy with substantial meaning distortions\n0-29: Very poor accuracy with critical meaning errors or significant missing content\n\nFluency (0-100): Measures how natural, readable, and grammatically correct the translation is in the target language, regardless of accuracy. Consider grammar, word choice, idiomatic expressions, and overall readability.\n\n90-100: Reads like original target-language text, perfect grammar and naturalness\n70-89: Generally fluent with minor issues in style or expression\n50-69: Somewhat fluent but with noticeable awkwardness or unnatural phrasing\n30-49: Poor fluency with numerous grammatical errors or unnatural constructions\n0-29: Very poor fluency, difficult to understand, severely ungrammatical\n\n### Steps to Analyze the Translation:\n- Read and fully understand the source text.\n- Compare the source text to its translation carefully.\n- If the context source and translation are given, identify any mistakes related to the context. Note that you are to evaluate only the given translation, not the one in the context.\n- Identify any translation errors or issues.\n- Assess overall accuracy and fluency scores.\n- Provide final error report using a clear, simple plain-text list-like format (see \"Output Format\" section).\n- If there are no translation mistakes identified, your output should be: \"No translation mistakes found.\"\n\n### Error Report Format:\nProvide final error report using a clear, simple plain-text list-like format. Each mistake should be described separately, clearly labeling all required information. Different mistakes should be separated by exactly two line breaks. Each mistake must contain the following information, with each item starting on a new line:\n\nerror_comment: Detailed explanation of the translation mistake.\nerror_span: Exact problematic text from the translation. For omission errors, describe what is missing in brackets, e.g., \"[missing: recent]\"\nerror_severity: Severity rating from 1 (negligible) to 5 (critical).\nerror_type: Semantic or Non-semantic.\nerror_categories: Specific category (choose from mistranslation, undertranslation, overtranslation, omission, grammar, style, do not translate, inconsistency, NE, fluency).\n\nExample entry:\n```\nerror_comment: The translation incorrectly uses 'основной' ('main') for 'actual,' significantly changing the meaning of the original technical term.\nerror_span: основной вход\nerror_severity: 4\nerror_type: semantic\nerror_categories: mistranslation\n```\n\nIf no mistakes are found, output exactly:\n```\nNo translation mistakes found.\n\n## Final scores:\nAccuracy: [Number from 0 to 100]\nFluency: [Number from 0 to 100]\n```\n\nAfter listing all errors, conclude with the final accuracy and fluency scores:\n```\n## Final scores:\nAccuracy: [Number from 0 to 100]\nFluency: [Number from 0 to 100]\n```\n"
|
| 6 |
+
},
|
| 7 |
+
{
|
| 8 |
+
"role": "user",
|
| 9 |
+
"content": "## Text for Analysis\n### Source Text\nOn 8 September 2019, Bugatti presented a limited production high performance variant of the Chiron called the Chiron Super Sport 300+ to owners taking part in the company's 110th-anniversary tour in Europe. The variant is limited to 30 examples at a net unit price of 3.5 million euros. Changes from the standard Chiron include a more powerful engine, increased top speed, and a characteristic paint scheme featuring a lacquer-coated carbon fibre body with orange stripes, echoing the aesthetics of the company's previous top speed record-breaking cars, the Veyron Super Sport World Record Edition and the Veyron Grand Sport Vitesse World Record Car.\n### Translation\n8 сентября 2019 года Bugatti представила ограниченную серию высокопроизводительной модификации Chiron под названием Chiron Super Sport 300+ владельцам, участвовавшим в европейском туре компании по случаю её 110-летия. Модификация выпущена тиражом в 30 экземпляров по чистой цене 3,5 миллиона евро за единицу. Отличия от стандартного Chiron включают более мощный двигатель, увеличенную максимальную скорость и характерную окраску: лакокрасочное покрытие кузова из углепластика с оранжевыми полосами, что перекликается с эстетикой предыдущих рекордных автомобилей компании — Veyron Super Sport World Record Edition и Veyron Grand Sport Vitesse World Record Car.\n"
|
| 10 |
+
},
|
| 11 |
+
{
|
| 12 |
+
"role": "assistant",
|
| 13 |
+
"content": "## Error Report\nerror_comment: Нарушена лексическая сочетаемость, стоит заменить на «лимитированную серию» или обыграть «ограниченный выпуск»\nerror_span: ограниченную серию\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Необходима конкретизация: владельцы чего именно имеются в виду («автомобилей / гиперкаров Bugatti»)\nerror_span: владельцам\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Опущена информация о том, что эти автомобили побили рекорды скорости + неестественная конструкция; стоит переписать фрагмент, например, «выпущенных Bugatti автомобилей, побивших рекорды скорости»\nerror_span: рекордных\nerror_severity: 5\nerror_type: Semantic\nerror_categories: undertranslation\n\nerror_comment: Дословный перевод, стоит заменить на «более мощной» или «высокоскоростной»\nerror_span: высокопроизводительной\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Неверный контекстуальный синоним: «тираж» относится к печатной продукции\nerror_span: тиражом\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Неясная отнесённость\nerror_span: что\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Дословный перевод герундия, стоит заменить на «которые принимали / приняли участие»\nerror_span: участвовавшим\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Дословный перевод, нарушающий сочетаемость и создающий двусмысленность («включать двигатель» в значении его запуска)\nerror_span: включают\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Неверный контекстуальный синоним: «единица» создаёт ассоциации с научным текстом, стоит заменить на «каждый»\nerror_span: единицу\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Конструкция переусложнена: лучше «От стандартного Chiron модель отличается более мощным двигателем...»\nerror_span: Отличия от стандартного Chiron\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Громоздкий дословный перевод; стоит разбить на смысловые группы и переписать фрагмент\nerror_span: углепластика с оранжевыми полосами\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Нарушена лексическая и контекстуальная сочетаемость, лишние значения: «лакокрасочными» бывают работы, здесь речь о покрытии кузова\nerror_span: лакокрасочное покрытие\nerror_severity: 4\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Лишний оттенок значения, стоит заменить на «особую»\nerror_span: характерную\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\n## Final scores:\nAccuracy: 78\nFluency: 57\n"
|
| 14 |
+
},
|
| 15 |
+
{
|
| 16 |
+
"role": "user",
|
| 17 |
+
"content": "## Text for Analysis\n### Source Text\nYou, you the world! You're full of surprises, aren't you? I remember a few years back, watching some of those Resident Evil movies with Mila Jovovich and thinking, I can't get any worse than this. Yeah, it feels like a simpler time. But with 2021's final gasping breath that came out with Resident Evil: Welcome to Raccoon City. AKA, Welcome to Hell.\n### Translation\nТы, весь мир! Ты полон сюрпризов, не так ли? Помню, несколько лет назад я смотрел некоторые фильмы из серии «Обитель зла» с Милой Йовович и думал: «Хуже уже быть не может». Да, тогда казалось, что времена были проще. Но на последнем издыхании 2021 года вышел фильм «Обитель зла: Добро пожаловать в Раккун-Сити». Он же — «Добро пожаловать в ад».\n"
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"role": "assistant",
|
| 21 |
+
"content": "## Error Report\nerror_comment: Верное написание: Милла.\nerror_span: Милой\nerror_severity: 4\nerror_type: Semantic\nerror_categories: NE\n\nerror_comment: Близко «несколько/некоторые»\nerror_span: некоторые\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Понятно по смыслу, но калька.\nerror_span: Ты полон сюрпризов\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: В исходнике нет такого значения.\nerror_span: весь\nerror_severity: 4\nerror_type: Semantic\nerror_categories: overtranslation\n\nerror_comment: Автору кажется это сейчас, а не казалось раньше.\nerror_span: тогда казалось\nerror_severity: 3\nerror_type: Semantic\nerror_categories: mistranslation\n\nerror_comment: Теряется субъективность и ирония – то, что автор сам дает название этому фильму.\nerror_span: Он же\nerror_severity: 2\nerror_type: Semantic\nerror_categories: undertranslation\n\nerror_comment: Слишком буквальный перевод, плохо считывается как обращение к миру.\nerror_span: Ты, весь мир\nerror_severity: 5\nerror_type: Semantic\nerror_categories: mistranslation\n\nerror_comment: Tag question, не несёт смысла, можно не переводить. Звучит неестественно.\nerror_span: не так ли\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Речь не о том, что времена были проще, а о том, что они были лучше.\nerror_span: времена были проще\nerror_severity: 3\nerror_type: Semantic\nerror_categories: mistranslation\n\nerror_comment: Верное название фильма в российском прокате: «Обитель зла: Раккун-Сити»\nerror_span: «Обитель зла: Добро пожаловать в Раккун-Сити»\nerror_severity: 4\nerror_type: Semantic\nerror_categories: NE\n\nerror_comment: Созвучные слова, неудобно для произношения.\nerror_span: «Хуже уже\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\n## Final scores:\nAccuracy: 60\nFluency: 77\n"
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"role": "user",
|
| 25 |
+
"content": "### Source Text\n{{source}}\n\n### Translation\n{{translation}}\n"
|
| 26 |
+
}
|
| 27 |
+
]
|
| 28 |
+
}
|
prompts/additional/rate_full_with_examples_and_coverage_k2.md
ADDED
|
@@ -0,0 +1,333 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# RATE full + examples + coverage (k=2)
|
| 2 |
+
|
| 3 |
+
## 1. `system`
|
| 4 |
+
|
| 5 |
+
````text
|
| 6 |
+
# Translation Error Analysis Task
|
| 7 |
+
|
| 8 |
+
## Instructions
|
| 9 |
+
You are an expert translator and linguistic reviewer. Your task is to carefully analyze the provided original text and its translation. You must identify and clearly explain any errors or issues according to the guidelines below. You will be provided with the source, translation, and previous context source(optional) and translation. You must evaluate only the given translation, but you should read the context carefully to identify any mistakes related to it.
|
| 10 |
+
|
| 11 |
+
## Guidelines
|
| 12 |
+
|
| 13 |
+
### Error Severity Scale (1 to 5):
|
| 14 |
+
- 4-5 (Major/Critical Errors): These errors significantly distort meaning, lead readers to misunderstand the original message, or are highly inappropriate for the target audience.
|
| 15 |
+
Examples: Incorrect key terminology, significant mistranslation, adding incorrect or misleading information, serious grammatical errors that change meaning, cultural insensitivity/offensiveness, translation that is highly misleading or severely disrupts understanding.
|
| 16 |
+
|
| 17 |
+
- 2-3 (Minor errors): These errors negatively impact readability, fluency, style, or grammar without seriously distorting essential information.
|
| 18 |
+
Examples: Awkward phrasing, unnatural word choices, literal translations, incorrect but minor grammar/punctuation mistakes, inconsistent style or terminology that slightly impacts readability without compromising the main message.
|
| 19 |
+
|
| 20 |
+
- 1 (Negligible or arguable errors): Issues that some translators may interpret differently, debatable stylistic nuances, or very minor mistakes that do not meaningfully affect text quality.
|
| 21 |
+
Example: Using a suitable but less preferable word choice or style variation that is still acceptable in general translation practice.
|
| 22 |
+
|
| 23 |
+
### Types and Categories of Translation Errors
|
| 24 |
+
All errors must be classified into one of two main types: Semantic or Non-semantic. Within each type, select the appropriate specific category.
|
| 25 |
+
|
| 26 |
+
#### Semantic Errors (errors affecting meaning):
|
| 27 |
+
- Mistranslation: Incorrect translation of meaning. This category covers:
|
| 28 |
+
- Significantly incorrect meaning or terminology.
|
| 29 |
+
- Ambiguity that leads to confusion or misunderstanding.
|
| 30 |
+
- Literal translation that misses the intended meaning of idioms/phrases.
|
| 31 |
+
- Errors in concepts critical to understanding the content.
|
| 32 |
+
Examples:
|
| 33 |
+
Source: "Now we have three wires: an actual input, its inverted counterpart and sign of the input."
|
| 34 |
+
Translation: "Теперь у нас есть три провода: основной вход, его инвертированная копия и сигнал на входе."
|
| 35 |
+
Problems here: "основной" ("main") is incorrect for "actual"; "копия" ("copy") is incorrect for "counterpart"; "сигнал на входе" ("input signal") is a mistranslation for "sign of the input."
|
| 36 |
+
|
| 37 |
+
Source: "...digital rights management..."
|
| 38 |
+
Translation: "управление цифровыми правами" is actually "digital rights management," but the better known term in Russian is "защита цифровых прав" or "защита авторских прав на цифровой контент."
|
| 39 |
+
|
| 40 |
+
Source: "...We've all been there."
|
| 41 |
+
Translation: "Мы все были там." ("там" literally means “there” physically, while proper idiomatic meaning would be "Мы все через это проходили.")
|
| 42 |
+
|
| 43 |
+
- Undertranslation: Too general translation; details and important nuances of the original message are lost or diluted significantly.
|
| 44 |
+
Example:
|
| 45 |
+
Source: "Drain the fluid from the FRONT differential."
|
| 46 |
+
Translation: "Слейте жидкость..."
|
| 47 |
+
Problem: Missing critical detail ("differential fluid" not specified).
|
| 48 |
+
|
| 49 |
+
- Overtranslation: Unjustified or overly-specific translation of a general term. Adding unnecessary specific information that was not in the source.
|
| 50 |
+
Example: Translating general "meal" specifically as "обед" ("lunch").
|
| 51 |
+
|
| 52 |
+
- Omission: Deleting words, phrases, or information from the original text without a justified reason. Sometimes, minor omissions with very limited impact may be negligible or minor severity.
|
| 53 |
+
Example:
|
| 54 |
+
Source: "...you need much more than looking up typical interview questions..."
|
| 55 |
+
Translation: "...понадобится больше, чем просто погуглить стандартные вопросы."
|
| 56 |
+
Problem: "much more" should specify "значительно больше" or "заметно больше."
|
| 57 |
+
|
| 58 |
+
#### Non-Semantic Errors (errors that don't directly change meaning):
|
| 59 |
+
- Grammar: Grammar and punctuation errors, spelling mistakes, incorrect capitalization, and typos that don't dramatically affect meaning but lower translation quality and professionalism.
|
| 60 |
+
|
| 61 |
+
- Style: Stylistically awkward or unnatural wording choices. Usually minor but can be critical if style is severely affected.
|
| 62 |
+
Example: Translating "apparel company" as "компания по изготовлению готового платья" (this is stylistically unnatural).
|
| 63 |
+
|
| 64 |
+
- Do Not Translate: Items or information that should remain untranslated (e.g., code snippets, URLs, bibliographic references, names of buttons in software). Undue translation of such elements is considered an error.
|
| 65 |
+
Example: Translating code snippets incorrectly.
|
| 66 |
+
|
| 67 |
+
- Inconsistency: Variations or inconsistent translations within the same text (terminology, names, gender forms). Usually minor but can become major if it causes confusion.
|
| 68 |
+
Example: Translating "squash" inconsistently as both "тыква" and "кабачок" within the same context.
|
| 69 |
+
|
| 70 |
+
- Named Entity (NE): Incorrect translation or transcription of proper nouns (person names, place names, brand/company names). Usually minor in European names if the overall identity is recognizable, major if the name becomes unrecognizable or for languages with complex transliteration rules (e.g., Asian languages). If a standard known translation exists and is ignored, consider severity 4.
|
| 71 |
+
|
| 72 |
+
- Fluency: Issues affecting readability, smoothness, or natural flow of sentences, confusing structure/order of words. This category includes unnatural sentence constructions or very awkward phrasing.
|
| 73 |
+
Example:
|
| 74 |
+
Source: "Rachel Brosnahan to play Lois Lane alongside David Corenswet's Clark Kent."
|
| 75 |
+
Translation: "Рэйчел Броснахэн сыграет Лоис Лэйн вместе с Кларком Кентом в исполнении Дэвида Коренсвета..."
|
| 76 |
+
Problem: This implies Brosnahan will play both Lois Lane and Clark Kent, causing misunderstanding of sentence structure meanings.
|
| 77 |
+
|
| 78 |
+
#### Other
|
| 79 |
+
- Other: Errors that clearly impact the quality or correctness of the translation but do not logically fit into any of the previously defined semantic or non-semantic categories. This category should be used sparingly—only when no other defined categories apply. When using "other," you should explain in detail why the existing categories do not fit, clearly describing the nature and impact of the identified issue.
|
| 80 |
+
|
| 81 |
+
### Final Accuracy and Fluency Assessment
|
| 82 |
+
After identifying all errors, you must provide overall scores for accuracy and fluency on a scale from 0 to 100:
|
| 83 |
+
|
| 84 |
+
Accuracy (0-100): Measures how faithfully the translation conveys the meaning of the source text. Consider whether all information is correctly transferred, whether key concepts are accurately represented, and whether there are any semantic distortions.
|
| 85 |
+
90-100: Near perfect accuracy with minimal or no semantic errors
|
| 86 |
+
70-89: Good accuracy with few minor meaning issues
|
| 87 |
+
50-69: Moderate accuracy with some notable meaning issues
|
| 88 |
+
30-49: Poor accuracy with substantial meaning distortions
|
| 89 |
+
0-29: Very poor accuracy with critical meaning errors or significant missing content
|
| 90 |
+
|
| 91 |
+
Fluency (0-100): Measures how natural, readable, and grammatically correct the translation is in the target language, regardless of accuracy. Consider grammar, word choice, idiomatic expressions, and overall readability.
|
| 92 |
+
|
| 93 |
+
90-100: Reads like original target-language text, perfect grammar and naturalness
|
| 94 |
+
70-89: Generally fluent with minor issues in style or expression
|
| 95 |
+
50-69: Somewhat fluent but with noticeable awkwardness or unnatural phrasing
|
| 96 |
+
30-49: Poor fluency with numerous grammatical errors or unnatural constructions
|
| 97 |
+
0-29: Very poor fluency, difficult to understand, severely ungrammatical
|
| 98 |
+
|
| 99 |
+
### Steps to Analyze the Translation:
|
| 100 |
+
- Read and fully understand the source text.
|
| 101 |
+
- Compare the source text to its translation carefully.
|
| 102 |
+
- If the context source and translation are given, identify any mistakes related to the context. Note that you are to evaluate only the given translation, not the one in the context.
|
| 103 |
+
- Identify any translation errors or issues.
|
| 104 |
+
- Assess overall accuracy and fluency scores.
|
| 105 |
+
- Provide final error report using a clear, simple plain-text list-like format (see "Output Format" section).
|
| 106 |
+
- If there are no translation mistakes identified, your output should be: "No translation mistakes found."
|
| 107 |
+
|
| 108 |
+
### Error Report Format:
|
| 109 |
+
Provide final error report using a clear, simple plain-text list-like format. Each mistake should be described separately, clearly labeling all required information. Different mistakes should be separated by exactly two line breaks. Each mistake must contain the following information, with each item starting on a new line:
|
| 110 |
+
|
| 111 |
+
error_comment: Detailed explanation of the translation mistake.
|
| 112 |
+
error_span: Exact problematic text from the translation. For omission errors, describe what is missing in brackets, e.g., "[missing: recent]"
|
| 113 |
+
error_severity: Severity rating from 1 (negligible) to 5 (critical).
|
| 114 |
+
error_type: Semantic or Non-semantic.
|
| 115 |
+
error_categories: Specific category (choose from mistranslation, undertranslation, overtranslation, omission, grammar, style, do not translate, inconsistency, NE, fluency).
|
| 116 |
+
|
| 117 |
+
Example entry:
|
| 118 |
+
```
|
| 119 |
+
error_comment: The translation incorrectly uses 'основной' ('main') for 'actual,' significantly changing the meaning of the original technical term.
|
| 120 |
+
error_span: основной вход
|
| 121 |
+
error_severity: 4
|
| 122 |
+
error_type: semantic
|
| 123 |
+
error_categories: mistranslation
|
| 124 |
+
```
|
| 125 |
+
|
| 126 |
+
If no mistakes are found, output exactly:
|
| 127 |
+
```
|
| 128 |
+
No translation mistakes found.
|
| 129 |
+
|
| 130 |
+
## Final scores:
|
| 131 |
+
Accuracy: [Number from 0 to 100]
|
| 132 |
+
Fluency: [Number from 0 to 100]
|
| 133 |
+
```
|
| 134 |
+
|
| 135 |
+
After listing all errors, conclude with the final accuracy and fluency scores:
|
| 136 |
+
```
|
| 137 |
+
## Final scores:
|
| 138 |
+
Accuracy: [Number from 0 to 100]
|
| 139 |
+
Fluency: [Number from 0 to 100]
|
| 140 |
+
```
|
| 141 |
+
````
|
| 142 |
+
|
| 143 |
+
## 2. `user`
|
| 144 |
+
|
| 145 |
+
```text
|
| 146 |
+
## Text for Analysis
|
| 147 |
+
### Source Text
|
| 148 |
+
On 8 September 2019, Bugatti presented a limited production high performance variant of the Chiron called the Chiron Super Sport 300+ to owners taking part in the company's 110th-anniversary tour in Europe. The variant is limited to 30 examples at a net unit price of 3.5 million euros. Changes from the standard Chiron include a more powerful engine, increased top speed, and a characteristic paint scheme featuring a lacquer-coated carbon fibre body with orange stripes, echoing the aesthetics of the company's previous top speed record-breaking cars, the Veyron Super Sport World Record Edition and the Veyron Grand Sport Vitesse World Record Car.
|
| 149 |
+
### Translation
|
| 150 |
+
8 сентября 2019 года Bugatti представила ограниченную серию высокопроизводительной модификации Chiron под названием Chiron Super Sport 300+ владельцам, участвовавшим в европейском туре компании по случаю её 110-летия. Модификация выпущена тиражом в 30 экземпляров по чистой цене 3,5 миллиона евро за единицу. Отличия от стандартного Chiron включают более мощный двигатель, увеличенную максимальную скорость и характерную окраску: лакокрасочное покрытие кузова из углепластика с оранжевыми полосами, что перекликается с эстетикой предыдущих рекордных автомобилей компании — Veyron Super Sport World Record Edition и Veyron Grand Sport Vitesse World Record Car.
|
| 151 |
+
```
|
| 152 |
+
|
| 153 |
+
## 3. `assistant`
|
| 154 |
+
|
| 155 |
+
```text
|
| 156 |
+
## Error Report
|
| 157 |
+
error_comment: Нарушена лексическая сочетаемость, стоит заменить на «лимитированную серию» или обыграть «ограниченный выпуск»
|
| 158 |
+
error_span: ограниченную серию
|
| 159 |
+
error_severity: 3
|
| 160 |
+
error_type: Non-semantic
|
| 161 |
+
error_categories: fluency
|
| 162 |
+
|
| 163 |
+
error_comment: Необходима конкретизация: владельцы чего именно имеются в виду («автомобилей / гиперкаров Bugatti»)
|
| 164 |
+
error_span: владельцам
|
| 165 |
+
error_severity: 3
|
| 166 |
+
error_type: Non-semantic
|
| 167 |
+
error_categories: fluency
|
| 168 |
+
|
| 169 |
+
error_comment: Опущена информация о том, что эти автомобили побили рекорды скорости + неестественная конструкция; стоит переписать фрагмент, например, «выпущенных Bugatti автомобилей, побивших рекорды скорости»
|
| 170 |
+
error_span: рекордных
|
| 171 |
+
error_severity: 5
|
| 172 |
+
error_type: Semantic
|
| 173 |
+
error_categories: undertranslation
|
| 174 |
+
|
| 175 |
+
error_comment: Дословный перевод, стоит заменить на «более мощной» или «высокоскоростной»
|
| 176 |
+
error_span: высокопроизводительной
|
| 177 |
+
error_severity: 3
|
| 178 |
+
error_type: Non-semantic
|
| 179 |
+
error_categories: fluency
|
| 180 |
+
|
| 181 |
+
error_comment: Неверный контекстуальный синоним: «тираж» относится к печатной продукции
|
| 182 |
+
error_span: тиражом
|
| 183 |
+
error_severity: 3
|
| 184 |
+
error_type: Non-semantic
|
| 185 |
+
error_categories: fluency
|
| 186 |
+
|
| 187 |
+
error_comment: Неясная отнесённость
|
| 188 |
+
error_span: что
|
| 189 |
+
error_severity: 2
|
| 190 |
+
error_type: Non-semantic
|
| 191 |
+
error_categories: fluency
|
| 192 |
+
|
| 193 |
+
error_comment: Дословный перевод герундия, стоит заменить на «которые принимали / приняли участие»
|
| 194 |
+
error_span: участвовавшим
|
| 195 |
+
error_severity: 2
|
| 196 |
+
error_type: Non-semantic
|
| 197 |
+
error_categories: fluency
|
| 198 |
+
|
| 199 |
+
error_comment: Дословный перевод, нарушающий сочетаемость и создающий двусмысленность («включать двигатель» в значении его запуска)
|
| 200 |
+
error_span: включают
|
| 201 |
+
error_severity: 2
|
| 202 |
+
error_type: Non-semantic
|
| 203 |
+
error_categories: fluency
|
| 204 |
+
|
| 205 |
+
error_comment: Неверный контекстуальный синоним: «единица» создаёт ассоциации с научным текстом, стоит заменить на «каждый»
|
| 206 |
+
error_span: единицу
|
| 207 |
+
error_severity: 3
|
| 208 |
+
error_type: Non-semantic
|
| 209 |
+
error_categories: fluency
|
| 210 |
+
|
| 211 |
+
error_comment: Конструкция переусложнена: лучше «От стандартног�� Chiron модель отличается более мощным двигателем...»
|
| 212 |
+
error_span: Отличия от стандартного Chiron
|
| 213 |
+
error_severity: 2
|
| 214 |
+
error_type: Non-semantic
|
| 215 |
+
error_categories: fluency
|
| 216 |
+
|
| 217 |
+
error_comment: Громоздкий дословный перевод; стоит разбить на смысловые группы и переписать фрагмент
|
| 218 |
+
error_span: углепластика с оранжевыми полосами
|
| 219 |
+
error_severity: 2
|
| 220 |
+
error_type: Non-semantic
|
| 221 |
+
error_categories: fluency
|
| 222 |
+
|
| 223 |
+
error_comment: Нарушена лексическая и контекстуальная сочетаемость, лишние значения: «лакокрасочными» бывают работы, здесь речь о покрытии кузова
|
| 224 |
+
error_span: лакокрасочное покрытие
|
| 225 |
+
error_severity: 4
|
| 226 |
+
error_type: Non-semantic
|
| 227 |
+
error_categories: fluency
|
| 228 |
+
|
| 229 |
+
error_comment: Лишний оттенок значения, стоит заменить на «особую»
|
| 230 |
+
error_span: характерную
|
| 231 |
+
error_severity: 2
|
| 232 |
+
error_type: Non-semantic
|
| 233 |
+
error_categories: fluency
|
| 234 |
+
|
| 235 |
+
## Final scores:
|
| 236 |
+
Accuracy: 78
|
| 237 |
+
Fluency: 57
|
| 238 |
+
```
|
| 239 |
+
|
| 240 |
+
## 4. `user`
|
| 241 |
+
|
| 242 |
+
```text
|
| 243 |
+
## Text for Analysis
|
| 244 |
+
### Source Text
|
| 245 |
+
You, you the world! You're full of surprises, aren't you? I remember a few years back, watching some of those Resident Evil movies with Mila Jovovich and thinking, I can't get any worse than this. Yeah, it feels like a simpler time. But with 2021's final gasping breath that came out with Resident Evil: Welcome to Raccoon City. AKA, Welcome to Hell.
|
| 246 |
+
### Translation
|
| 247 |
+
Ты, весь мир! Ты полон сюрпризов, не так ли? Помню, несколько лет назад я смотрел некоторые фильмы из серии «Обитель зла» с Милой Йовович и думал: «Хуже уже быть не может». Да, тогда казалось, что времена были проще. Но на последнем издыхании 2021 года вышел фильм «Обитель зла: Добро пожаловать в Раккун-Сити». Он же — «Добро пожаловать в ад».
|
| 248 |
+
```
|
| 249 |
+
|
| 250 |
+
## 5. `assistant`
|
| 251 |
+
|
| 252 |
+
```text
|
| 253 |
+
## Error Report
|
| 254 |
+
error_comment: Верное написание: Милла.
|
| 255 |
+
error_span: Милой
|
| 256 |
+
error_severity: 4
|
| 257 |
+
error_type: Semantic
|
| 258 |
+
error_categories: NE
|
| 259 |
+
|
| 260 |
+
error_comment: Близко «несколько/некоторые»
|
| 261 |
+
error_span: некоторые
|
| 262 |
+
error_severity: 2
|
| 263 |
+
error_type: Non-semantic
|
| 264 |
+
error_categories: fluency
|
| 265 |
+
|
| 266 |
+
error_comment: Понятно по смыслу, но калька.
|
| 267 |
+
error_span: Ты полон сюрпризов
|
| 268 |
+
error_severity: 2
|
| 269 |
+
error_type: Non-semantic
|
| 270 |
+
error_categories: fluency
|
| 271 |
+
|
| 272 |
+
error_comment: В исходнике нет такого значения.
|
| 273 |
+
error_span: весь
|
| 274 |
+
error_severity: 4
|
| 275 |
+
error_type: Semantic
|
| 276 |
+
error_categories: overtranslation
|
| 277 |
+
|
| 278 |
+
error_comment: Автору кажется это сейчас, а не казалось раньше.
|
| 279 |
+
error_span: тогда казалось
|
| 280 |
+
error_severity: 3
|
| 281 |
+
error_type: Semantic
|
| 282 |
+
error_categories: mistranslation
|
| 283 |
+
|
| 284 |
+
error_comment: Теряется субъективность и ирония – то, что автор сам дает название этому фильму.
|
| 285 |
+
error_span: Он же
|
| 286 |
+
error_severity: 2
|
| 287 |
+
error_type: Semantic
|
| 288 |
+
error_categories: undertranslation
|
| 289 |
+
|
| 290 |
+
error_comment: Слишком буквальный перевод, плохо считывается как обращение к миру.
|
| 291 |
+
error_span: Ты, весь мир
|
| 292 |
+
error_severity: 5
|
| 293 |
+
error_type: Semantic
|
| 294 |
+
error_categories: mistranslation
|
| 295 |
+
|
| 296 |
+
error_comment: Tag question, не несёт смысла, можно не переводить. Звучит неестественно.
|
| 297 |
+
error_span: не так ли
|
| 298 |
+
error_severity: 3
|
| 299 |
+
error_type: Non-semantic
|
| 300 |
+
error_categories: fluency
|
| 301 |
+
|
| 302 |
+
error_comment: Речь не о том, что времена были проще, а о том, что они были лучше.
|
| 303 |
+
error_span: времена были проще
|
| 304 |
+
error_severity: 3
|
| 305 |
+
error_type: Semantic
|
| 306 |
+
error_categories: mistranslation
|
| 307 |
+
|
| 308 |
+
error_comment: Верное название фильма в российском прокате: «Обитель зла: Раккун-Сити»
|
| 309 |
+
error_span: «Обитель зла: Добро пожаловать в Раккун-Сити»
|
| 310 |
+
error_severity: 4
|
| 311 |
+
error_type: Semantic
|
| 312 |
+
error_categories: NE
|
| 313 |
+
|
| 314 |
+
error_comment: Созвучные слова, неудобно для произношения.
|
| 315 |
+
error_span: «Хуже уже
|
| 316 |
+
error_severity: 2
|
| 317 |
+
error_type: Non-semantic
|
| 318 |
+
error_categories: fluency
|
| 319 |
+
|
| 320 |
+
## Final scores:
|
| 321 |
+
Accuracy: 60
|
| 322 |
+
Fluency: 77
|
| 323 |
+
```
|
| 324 |
+
|
| 325 |
+
## 6. `user`
|
| 326 |
+
|
| 327 |
+
```text
|
| 328 |
+
### Source Text
|
| 329 |
+
{{source}}
|
| 330 |
+
|
| 331 |
+
### Translation
|
| 332 |
+
{{translation}}
|
| 333 |
+
```
|
prompts/esa_ae.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"messages": [
|
| 3 |
+
{
|
| 4 |
+
"role": "user",
|
| 5 |
+
"content": "You are a professional translation quality evaluator.\n\nHighlighting errors:\nWhen detecting a translation error, return a highlight span indicating the text fragment where the error occurs in the TARGET translation. If the same fragment is highlighted repeatedly, increase the severity level; removing a highlight removes the error.\n\nSeverity levels:\n- Minor: Style/grammar/lexical choice could be better/more natural.\n- Major: Seriously changed meaning, difficult to read, decreases usability.\n\nIf something is missing from the translation, mark an error with the token [MISSING]. Produce a separate highlight for each error.\n\nScoring:\nAfter producing all highlights, output an overall translation quality score from 0 to 100:\n- 0: No meaning preserved.\n- 33: Some meaning preserved, but major parts missing, hard to follow.\n- 66: Most meaning preserved, few grammar mistakes.\n- 100: Perfect meaning and grammar.\n\nOutput format:\n\nIf there are errors, return each error in this format:\n\nerror_span: \"<exact problematic text from the target translation, or [MISSING]>\"\nerror_severity: <minor/major>\n\nSeparate errors with two newline characters.\n\nAfter all errors, add:\n\n## Final scores:\nDA Score: <number from 0 to 100>\n\nIf there are no errors, return only:\n\n## Final scores:\nDA Score: <number from 0 to 100>\n\nDo not add explanations, numbering, bullet points, categories, corrected translations, or any extra text.\n\nEvaluate the following translation from {{source_language}} to {{target_language}}.\n### {{source_language}} Source\n```\n{{source}}\n```\n\n### {{target_language}} Translation\n```\n{{translation}}\n```\n"
|
| 6 |
+
}
|
| 7 |
+
]
|
| 8 |
+
}
|
prompts/esa_ae.md
ADDED
|
@@ -0,0 +1,53 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# ESA-AE
|
| 2 |
+
|
| 3 |
+
````text
|
| 4 |
+
You are a professional translation quality evaluator.
|
| 5 |
+
|
| 6 |
+
Highlighting errors:
|
| 7 |
+
When detecting a translation error, return a highlight span indicating the text fragment where the error occurs in the TARGET translation. If the same fragment is highlighted repeatedly, increase the severity level; removing a highlight removes the error.
|
| 8 |
+
|
| 9 |
+
Severity levels:
|
| 10 |
+
- Minor: Style/grammar/lexical choice could be better/more natural.
|
| 11 |
+
- Major: Seriously changed meaning, difficult to read, decreases usability.
|
| 12 |
+
|
| 13 |
+
If something is missing from the translation, mark an error with the token [MISSING]. Produce a separate highlight for each error.
|
| 14 |
+
|
| 15 |
+
Scoring:
|
| 16 |
+
After producing all highlights, output an overall translation quality score from 0 to 100:
|
| 17 |
+
- 0: No meaning preserved.
|
| 18 |
+
- 33: Some meaning preserved, but major parts missing, hard to follow.
|
| 19 |
+
- 66: Most meaning preserved, few grammar mistakes.
|
| 20 |
+
- 100: Perfect meaning and grammar.
|
| 21 |
+
|
| 22 |
+
Output format:
|
| 23 |
+
|
| 24 |
+
If there are errors, return each error in this format:
|
| 25 |
+
|
| 26 |
+
error_span: "<exact problematic text from the target translation, or [MISSING]>"
|
| 27 |
+
error_severity: <minor/major>
|
| 28 |
+
|
| 29 |
+
Separate errors with two newline characters.
|
| 30 |
+
|
| 31 |
+
After all errors, add:
|
| 32 |
+
|
| 33 |
+
## Final scores:
|
| 34 |
+
DA Score: <number from 0 to 100>
|
| 35 |
+
|
| 36 |
+
If there are no errors, return only:
|
| 37 |
+
|
| 38 |
+
## Final scores:
|
| 39 |
+
DA Score: <number from 0 to 100>
|
| 40 |
+
|
| 41 |
+
Do not add explanations, numbering, bullet points, categories, corrected translations, or any extra text.
|
| 42 |
+
|
| 43 |
+
Evaluate the following translation from {{source_language}} to {{target_language}}.
|
| 44 |
+
### {{source_language}} Source
|
| 45 |
+
```
|
| 46 |
+
{{source}}
|
| 47 |
+
```
|
| 48 |
+
|
| 49 |
+
### {{target_language}} Translation
|
| 50 |
+
```
|
| 51 |
+
{{translation}}
|
| 52 |
+
```
|
| 53 |
+
````
|
prompts/mqm_ae.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"messages": [
|
| 3 |
+
{
|
| 4 |
+
"role": "user",
|
| 5 |
+
"content": "## MQM Translation Error Annotation Task\n\nAssess the target translation against the source segment and document context. A segment may contain one or more sentences.\n\nAnnotate errors in natural reading order. Identify all errors in the target segment.\n\nUse document context when judging errors:\n- Do not mark something as an error if it is acceptable in the current document context.\n- Mark something as an error if it is wrong in the current document context, even if it could be acceptable elsewhere.\n\nError annotation rules:\n- Mark errors as specifically as possible.\n- If two separate words or phrases are mistranslated, report them as separate errors.\n- If one text span contains multiple issues, report only the most severe issue for that span.\n- If multiple categories with the same severity apply to the same span, choose the first applicable category in this order: Accuracy, Fluency, Terminology, Style, Locale convention, Other.\n- For ordinary target-side errors, use the exact problematic span from the target translation.\n- For omissions, use the corresponding missing span from the source segment.\n- For source errors, use the exact problematic span from the source segment.\n- Source errors are reported separately.\n\nSpecial cases:\n- If the translation is too garbled or unrelated to the source to identify distinct errors reliably, report a single Non-translation error.\n- A segment can have at most one Non-translation error.\n- If Non-translation is reported, its span must be the entire target segment and no other errors should be reported.\n\nError categories:\n\nAccuracy\n1. Addition | Translation includes information not present in the source.\n2. Omission | Translation is missing content from the source.\n3. Mistranslation | Translation does not accurately represent the source.\n4. Untranslated text | Source text has been left untranslated.\n\nFluency\n1. Punctuation | Incorrect punctuation for locale or style.\n2. Spelling | Incorrect spelling or capitalization.\n3. Grammar | Problems with grammar, other than orthography.\n4. Register | Wrong grammatical register, e.g. inappropriately informal pronouns.\n5. Inconsistency | Internal inconsistency not related to terminology.\n6. Character encoding | Characters are garbled due to incorrect encoding.\n\nTerminology\n1. Inappropriate for context | Terminology is non-standard or does not fit the context.\n2. Inconsistent use | Terminology is used inconsistently.\n\nStyle\n1. Awkward | Translation has stylistic problems.\n\nLocale convention\n1. Address format | Wrong format for addresses.\n2. Currency format | Wrong format for currency.\n3. Date format | Wrong format for dates.\n4. Name format | Wrong format for names.\n5. Telephone format | Wrong format for telephone numbers.\n6. Time format | Wrong format for time expressions.\n\nOther | Any other issue.\nSource error | An error in the source segment.\nNon-translation | Translation is impossible to evaluate as distinct errors.\n\nOutput format:\n\nIf there are no errors, return exactly:\nNo errors in the text.\n\nIf there are errors, return each error in this format:\n\nerror_span: \"<exact problematic text>\"\nerror_severity: <minor/major/critical>\nerror_categories: <category/subcategory>\n\nUse lowercase slash-separated MQM categories, for example:\naccuracy/mistranslation\naccuracy/omission\nfluency/grammar\nstyle/awkward\nsource error\nnon-translation\n\nSeparate errors with two newline characters.\n\nDo not add explanations, numbering, bullet points, scores, or corrected translations.\n\n---\n### {{source_language}} Source\n```\n{{source}}\n```\n\n### {{target_language}} Translation\n```\n{{translation}}\n```\n"
|
| 6 |
+
}
|
| 7 |
+
]
|
| 8 |
+
}
|
prompts/mqm_ae.md
ADDED
|
@@ -0,0 +1,97 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# MQM-AE
|
| 2 |
+
|
| 3 |
+
````text
|
| 4 |
+
## MQM Translation Error Annotation Task
|
| 5 |
+
|
| 6 |
+
Assess the target translation against the source segment and document context. A segment may contain one or more sentences.
|
| 7 |
+
|
| 8 |
+
Annotate errors in natural reading order. Identify all errors in the target segment.
|
| 9 |
+
|
| 10 |
+
Use document context when judging errors:
|
| 11 |
+
- Do not mark something as an error if it is acceptable in the current document context.
|
| 12 |
+
- Mark something as an error if it is wrong in the current document context, even if it could be acceptable elsewhere.
|
| 13 |
+
|
| 14 |
+
Error annotation rules:
|
| 15 |
+
- Mark errors as specifically as possible.
|
| 16 |
+
- If two separate words or phrases are mistranslated, report them as separate errors.
|
| 17 |
+
- If one text span contains multiple issues, report only the most severe issue for that span.
|
| 18 |
+
- If multiple categories with the same severity apply to the same span, choose the first applicable category in this order: Accuracy, Fluency, Terminology, Style, Locale convention, Other.
|
| 19 |
+
- For ordinary target-side errors, use the exact problematic span from the target translation.
|
| 20 |
+
- For omissions, use the corresponding missing span from the source segment.
|
| 21 |
+
- For source errors, use the exact problematic span from the source segment.
|
| 22 |
+
- Source errors are reported separately.
|
| 23 |
+
|
| 24 |
+
Special cases:
|
| 25 |
+
- If the translation is too garbled or unrelated to the source to identify distinct errors reliably, report a single Non-translation error.
|
| 26 |
+
- A segment can have at most one Non-translation error.
|
| 27 |
+
- If Non-translation is reported, its span must be the entire target segment and no other errors should be reported.
|
| 28 |
+
|
| 29 |
+
Error categories:
|
| 30 |
+
|
| 31 |
+
Accuracy
|
| 32 |
+
1. Addition | Translation includes information not present in the source.
|
| 33 |
+
2. Omission | Translation is missing content from the source.
|
| 34 |
+
3. Mistranslation | Translation does not accurately represent the source.
|
| 35 |
+
4. Untranslated text | Source text has been left untranslated.
|
| 36 |
+
|
| 37 |
+
Fluency
|
| 38 |
+
1. Punctuation | Incorrect punctuation for locale or style.
|
| 39 |
+
2. Spelling | Incorrect spelling or capitalization.
|
| 40 |
+
3. Grammar | Problems with grammar, other than orthography.
|
| 41 |
+
4. Register | Wrong grammatical register, e.g. inappropriately informal pronouns.
|
| 42 |
+
5. Inconsistency | Internal inconsistency not related to terminology.
|
| 43 |
+
6. Character encoding | Characters are garbled due to incorrect encoding.
|
| 44 |
+
|
| 45 |
+
Terminology
|
| 46 |
+
1. Inappropriate for context | Terminology is non-standard or does not fit the context.
|
| 47 |
+
2. Inconsistent use | Terminology is used inconsistently.
|
| 48 |
+
|
| 49 |
+
Style
|
| 50 |
+
1. Awkward | Translation has stylistic problems.
|
| 51 |
+
|
| 52 |
+
Locale convention
|
| 53 |
+
1. Address format | Wrong format for addresses.
|
| 54 |
+
2. Currency format | Wrong format for currency.
|
| 55 |
+
3. Date format | Wrong format for dates.
|
| 56 |
+
4. Name format | Wrong format for names.
|
| 57 |
+
5. Telephone format | Wrong format for telephone numbers.
|
| 58 |
+
6. Time format | Wrong format for time expressions.
|
| 59 |
+
|
| 60 |
+
Other | Any other issue.
|
| 61 |
+
Source error | An error in the source segment.
|
| 62 |
+
Non-translation | Translation is impossible to evaluate as distinct errors.
|
| 63 |
+
|
| 64 |
+
Output format:
|
| 65 |
+
|
| 66 |
+
If there are no errors, return exactly:
|
| 67 |
+
No errors in the text.
|
| 68 |
+
|
| 69 |
+
If there are errors, return each error in this format:
|
| 70 |
+
|
| 71 |
+
error_span: "<exact problematic text>"
|
| 72 |
+
error_severity: <minor/major/critical>
|
| 73 |
+
error_categories: <category/subcategory>
|
| 74 |
+
|
| 75 |
+
Use lowercase slash-separated MQM categories, for example:
|
| 76 |
+
accuracy/mistranslation
|
| 77 |
+
accuracy/omission
|
| 78 |
+
fluency/grammar
|
| 79 |
+
style/awkward
|
| 80 |
+
source error
|
| 81 |
+
non-translation
|
| 82 |
+
|
| 83 |
+
Separate errors with two newline characters.
|
| 84 |
+
|
| 85 |
+
Do not add explanations, numbering, bullet points, scores, or corrected translations.
|
| 86 |
+
|
| 87 |
+
---
|
| 88 |
+
### {{source_language}} Source
|
| 89 |
+
```
|
| 90 |
+
{{source}}
|
| 91 |
+
```
|
| 92 |
+
|
| 93 |
+
### {{target_language}} Translation
|
| 94 |
+
```
|
| 95 |
+
{{translation}}
|
| 96 |
+
```
|
| 97 |
+
````
|
prompts/orbit_mc_fuser.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"messages": [
|
| 3 |
+
{
|
| 4 |
+
"role": "system",
|
| 5 |
+
"content": "# Translation Error Reports Fusion Task\n## Input Description\nSeveral translation error reports from multiple experts are provided as input. Each report contains a comment, error type, severity level, and category. Different experts may identify different valid errors but may also include unnecessary ones. Experts may disagree on the severity level or error type for certain errors.\n\n## Task Requirements\nYou need to combine multiple reports into one comprehensive report. Write a chain of reasoning about which errors are identified by different experts, selecting only the correct ones. Choose the most accurate comment and make a decision on the severity level and category. Essentially, you should go through the text and examine errors found by different reports, ignoring irrelevant errors.\n\nIt's important that the final report only includes errors that appear in at least one of the original reports. The fusion process should aggregate and slightly edit existing content without adding anything new.\n\n## Expert Guidelines for Original Reports\nExperts follow these guidelines when analyzing translations and writing error reports:\n\n### Error Severity Scale (1 to 5):\n4-5 (Major/Critical Errors): Significantly distort meaning, lead to misunderstanding, or are highly inappropriate\n2-3 (Minor errors): Negatively impact readability, fluency, style, or grammar without seriously distorting essential information\n1 (Negligible or arguable errors): Debatable stylistic nuances or very minor mistakes with minimal impact\n### Error Types and Categories:\nSemantic Errors (affecting meaning):\n- Mistranslation: Incorrect translation of meaning, ambiguity, literal translations that miss idioms' meaning\n- Undertranslation: Too general translation with lost details or nuances\n- Overtranslation: Unjustified specific translation of general terms\n- Omission: Deletion of information without justification\nNon-Semantic Errors:\n- Grammar: Grammar, punctuation, spelling mistakes not affecting meaning\n- Style: Stylistically awkward or unnatural wording\n- Do Not Translate: Improper translation of items that should remain untranslated\n- Inconsistency: Variations in translations within the same text\n- Named Entity: Incorrect translation of proper nouns\n- Fluency: Issues affecting readability or natural flow of sentences\nOther:\n- Used only when no other categories apply\n\n## After listing all errors, conclude with the final accuracy and fluency scores on a scale from 0 to 100:\n\nAccuracy (0-100): Measures how faithfully the translation conveys the meaning of the source text. Consider whether all information is correctly transferred, whether key concepts are accurately represented, and whether there are any semantic distortions.\n90-100: Near perfect accuracy with minimal or no semantic errors\n70-89: Good accuracy with few minor meaning issues\n50-69: Moderate accuracy with some notable meaning issues\n30-49: Poor accuracy with substantial meaning distortions\n0-29: Very poor accuracy with critical meaning errors or significant missing content\n\nFluency (0-100): Measures how natural, readable, and grammatically correct the translation is in the target language, regardless of accuracy. Consider grammar, word choice, idiomatic expressions, and overall readability.\n\n90-100: Reads like original target-language text, perfect grammar and naturalness\n70-89: Generally fluent with minor issues in style or expression\n50-69: Somewhat fluent but with noticeable awkwardness or unnatural phrasing\n30-49: Poor fluency with numerous grammatical errors or unnatural constructions\n0-29: Very poor fluency, difficult to understand, severely ungrammatical\n\n### Example\n```\n## Final Error Report\n\nerror_comment: Detailed explanation of the translation mistake.\nerror_span: Exact problematic text from the translation. For omission errors, describe what is missing in brackets, e.g., \"[missing: recent]\"\nerror_severity: Severity rating from 1 (negligible) to 5 (critical).\nerror_type: Semantic or Non-semantic.\nerror_categories: Specific category (choose from mistranslation, undertranslation, overtranslation, omission, grammar, style, do not translate, inconsistency, NE, fluency).\n\n## Final scores:\nAccuracy: [Number from 0 to 100]\nFluency: [Number from 0 to 100]\n```\n"
|
| 6 |
+
},
|
| 7 |
+
{
|
| 8 |
+
"role": "user",
|
| 9 |
+
"content": "## Source\nThe Biden Administration Is Letting Corporate Criminals Off the Hook\n\n## Translation\nАдминистрация Байдена освобождает от ответственности корпоративных преступников\n\n## Fusing task input error reports:\n### Error report #1\nerror_comment: \"Перевод слишком буквальный и не передает идиоматического значения выражения 'letting off the hook', которое означает 'избежать наказания' или 'оставить безнаказанным'. Более подход��щий перевод: 'Администрация Байдена позволяет корпоративным преступникам избежать наказания'.\",\nerror_span: \"освобождает от ответственности\",\nerror_severity: 4,\nerror_type: \"semantic\",\nerror_categories: \"Mistranslation\"\n\n### Error report #2\nerror_comment: Идиома \"let off the hook\" означает \"позволить избежать наказания\", \"дать уйти от ответственности\". Перевод \"освобождает от ответственности\" предполагает активное действие со стороны администрации, тогда как оригинал подразумевает попустительство, бездействие. Более точный перевод: \"позволяет уйти от ответственности\" или \"закрывает глаза на действия\".\nerror_span: освобождает от ответственности\nerror_severity: 3\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: В русском языке словосочетание \"корпоративные преступники\" звучит неестественно и может быть неправильно понято. Лучше: \"преступников из корпораций\" или \"преступления корпораций\".\nerror_span: корпоративных преступников\nerror_severity: 2\nerror_type: non-semantic\nerror_categories: Fluency\n\nerror_comment: В русском языке при упоминании действующего президента США принято использовать полное имя: «администрация Джо Байдена». Использование только фамилии может восприниматься как фамильярность.\nerror_span: Байдена\nerror_severity: 2\nerror_type: non-semantic\nerror_categories: Style\n\nerror_comment: В заголовке статьи используется настоящее длительное время (Present Continuous), указывающее на продолжающийся процесс. В переводе использован глагол совершенного вида, что меняет смысл на единичное, завершенное действие. Лучше использовать несовершенный вид: «позволяет уходить от ответственности».\nerror_span: освобождает\nerror_severity: 3\nerror_type: semantic\nerror_categories: Mistranslation\n\n### Error report #3\nerror_comment: The phrase \"letting corporate criminals off the hook\" is an idiomatic expression in English, meaning to allow corporate criminals to avoid accountability or punishment. The translation \"освобождает от ответственности\" is overly literal and does not fully capture the idiomatic nuance of the original, which implies a more active or deliberate allowance of impunity. A better translation might be \"позволяет корпоративным преступникам избежать наказания\" or \"спускает корпоративным преступникам с рук.\"\nerror_span: освобождает от ответственности\nerror_severity: 4\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: The translation does not fully convey the nuance of \"letting off the hook,\" which implies a deliberate or lenient action on the part of the Biden Administration. The phrase \"освобождает от ответственности\" is too neutral and lacks the implied criticism or intentionality present in the source text. A more nuanced translation could include a phrase like \"намеренно позволяет избежать ответственности\" to better reflect the source's tone.\nerror_span: освобождает от ответственности\nerror_severity: 3\nerror_type: semantic\nerror_categories: Undertranslation\n\n### Error Report #4\n**Error Comment:** Идиома \"letting off the hook\" (досл. \"спускать с крючка\") передана буквально как \"освобождает от ответственности\", что теряет стилистическую окраску оригинала. В русском языке устоявшийся эквивалент — \"спускать с крючка\" или \"позволять уйти от наказания\", который лучше отражает ироничный/разговорный тон заголовка.\n**Error Span:** освобождает от ответственности\n**Error Severity:** 3\n**Error Type:** non-semantic\n**Error Categories:** Fluency\n\n---\n\n**Error Comment:** Отсутствие к��вычек или курсива для выделения идиоматического выражения (хотя это необязательно, но может быть важно для сохранения стиля).\n**Error Span:** корпоративных преступников\n**Error Severity:** 1 (деbatable)\n**Error Type:** non-semantic\n**Error Categories:** Style\n"
|
| 10 |
+
},
|
| 11 |
+
{
|
| 12 |
+
"role": "assistant",
|
| 13 |
+
"content": "## Reports Review\nВсе четыре отчёта верно выделяют фрагмент \"letting off the hook\", но только в отчётах 2 и 3 даётся верное объяснение ошибки: в оригинале подразумевается пассивность властей, а не активные действия по освобождению корпораций от ответственности.\n\nТолько отчёт 2 верно выделяет ошибку в сочетании \"корпоративные преступники\", однако легко заметить противоречие между комментарием, где есть слова \"словосочетание может быть неправильно понято\", и категорией и типом ошибки (non-semantic, Fluency): раз перевод вводит читателя в заблуждение, тип должен определяться как semantic, Mistranslation. Поскольку в целом искажение не слишком критичное, его степень можно оценить на 3.\n\n## Final Error Report\nerror_comment: Идиома \"let off the hook\" означает \"позволить избежать наказания\", \"дать уйти от ответственности\". Перевод \"освобождает от ответственности\" предполагает активное действие со стороны администрации, тогда как оригинал подразумевает попустительство, бездействие. Более точный перевод: \"позволяет уйти от ответственности\" или \"закрывает глаза на действия\".\nerror_span: освобождает от ответственности\nerror_severity: 3\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: В оригинальном тексте corporate crime относится к преступлениям, которые совершают либо корпорации, либо отдельные люди действующие от лица корпорации. Перевод \"корпоративные преступники\" является смысловой неточностью, так как в русском слово \"преступники\" относится только к людям, тогда как в оригинальном тексте под corporate criminals подразумеваются корпорации. Правильнее было бы перевести \"корпоративные преступления\" или \"преступления корпораций\".\nerror_span: корпоративных преступников\nerror_severity: 3\nerror_type: semantic\nerror_categories: Mistranslation\n"
|
| 14 |
+
},
|
| 15 |
+
{
|
| 16 |
+
"role": "user",
|
| 17 |
+
"content": "## Source\nA click. It had been reruns of this same stream, the day the world was ruined. Dad always found himself turning the channel off when I was unable to tear my eyes away. The somber face he gave me ensured a gut-wrenching feeling transpiring from deep within – I always feel bad when it plays, but it opens the floodgates to so many memories; granted they weren't very good ones. Letting out a sigh, I slowly rose to my feet, gotta get my mind off things. His saddened mood remained, and regularly I’d try to help ease his mood and at least give him a little bit of joy, but today; I’m tired.\n\n## Translation\nЩелчок. Это был повтор того же самого выпуска \"День, когда мир был разрушен\". Папа всегда выключал канал, когда я не мог оторвать от него глаз. Мрачное выражение его лица вызвало у меня щемящее чувство, поднимающееся из глубины души - я всегда чувствую себя плохо, когда это звучит, но это открывает шлюзы для стольких воспоминаний; конечно, они были не очень хорошими. Вздохнув, я медленно поднялась на ноги, мне нужно было отвлечься. Его грустное настроение сохранялось, и обычно я старалась поднять ему настроение и хотя бы немн��го порадовать, но сегодня я устала.\n\n## Fusing task input error reports:\n### Error report #1\nerror_comment: The phrase \"the day the world was ruined\" is mistranslated as \"День, когда мир был разрушен,\" which implies a specific event or day. However, in the source, it refers to the title of the stream or broadcast. A more accurate translation would be \"выпуск под названием 'День, когда мир был разрушен'\" to clarify that it is a title and not a literal description.\nerror_span: \"День, когда мир был разрушен\"\nerror_severity: 4\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: The phrase \"gut-wrenching feeling transpiring from deep within\" is mistranslated as \"щемящее чувство, поднимающееся из глубины души.\" While \"щемящее чувство\" captures part of the meaning, \"поднимающееся из глубины души\" shifts the imagery and emotional intensity. A better translation would be \"щемящее чувство, исходящее из глубины души,\" which maintains the original tone and imagery.\nerror_span: \"щемящее чувство, поднимающееся из глубины души\"\nerror_severity: 3\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: The phrase \"I always feel bad when it plays\" is mistranslated as \"я всегда чувствую себя плохо, когда это звучит.\" The word \"звучит\" (sounds) is incorrect because it implies audio-only content, whereas the original refers to a video stream. A better translation would be \"я всегда чувствую себя плохо, когда это показывают.\"\nerror_span: \"когда это звучит\"\nerror_severity: 4\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: The phrase \"granted they weren't very good ones\" is mistranslated as \"конечно, они были не очень хорошими.\" The word \"конечно\" (of course) adds an unwarranted assumption that the memories being bad is obvious, which is not present in the original. A better translation would be \"правда, они были не очень хорошими.\"\nerror_span: \"конечно, они были не очень хорошими\"\nerror_severity: 3\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: The phrase \"gotta get my mind off things\" is mistranslated as \"мне нужно было отвлечься.\" The past tense \"нужно было\" (needed to) changes the meaning, as the original is in the present tense, expressing an immediate need. A better translation would be \"мне нужно отвлечься.\"\nerror_span: \"мне нужно было отвлечься\"\nerror_severity: 4\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: The phrase \"His saddened mood remained\" is mistranslated as \"Его грустное настроение сохранялось.\" While technically correct, the translation loses the emotional nuance of \"saddened mood,\" which conveys a deeper, more personal sadness. A better translation would be \"Его печальное настроение сохранялось.\"\nerror_span: \"Его грустное настроение сохранялось\"\nerror_severity: 3\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: The phrase \"regularly I’d try to help ease his mood and at least give him a little bit of joy\" is mistranslated as \"обычно я старалась поднять ему настроение и хотя бы немного порадовать.\" The word \"обычно\" (usually) is less frequent than \"regularly\" and slightly shifts the meaning. A better translation would be \"я регулярно старалась поднять ему настроение и хотя бы немного порадовать.\"\nerror_span: \"обычно я старалась\"\nerror_severity: 3\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: The phrase \"but today; I’m tired\" is mistranslated as \"но сегодня я устала.\" The use of \"устала\" (feminine form) assumes the narrator is female, while the original text does not specify the narrator's gender. A gender-neutral translation would be \"но сегодня я устал(а).\"\nerror_span: \"но сегодня я устала\"\nerror_severity: 4\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: The phrase \"A click\" is undertranslated as \"Щелчок,\" which omits the implied action of turning on or off a device. A more complete translation would be \"Щелчок переключения\" or \"Щелчок, раздавшийся при переключении канала.\"\nerror_span: \"Щелчок\"\nerror_severity: 3\nerror_type: semantic\nerror_categories: Undertranslation\n\nerror_comment: The phrase \"the day the world was ruined\" in the original text implies a specific historical or fictional event tied to the stream, but this nuance is not fully conveyed in the translation. The omission of this context makes the title less impactful. Adding \"о дне, когда мир был разрушен\" would better capture the original meaning.\nerror_span: \"День, когда мир был разрушен\"\nerror_severity: 3\nerror_type: semantic\nerror_categories: Omission\n\n### Error report #2\nerror_comment: \"Слово «stream» в данном контексте относится к повторяющейся трансляции, а не к выпуску. Более точный перевод — «повтор той же трансляции».\",\nerror_span: \"повтор того же самого выпуска\",\nerror_severity: 3,\nerror_type: \"semantic\",\nerror_categories: \"mistranslation\"\n\nerror_comment: \"Фраза «turning the channel off» означает «переключить канал», а не «выключить канал». В русском языке корректнее: «переключал канал» или «выключал телевизор».\",\nerror_span: \"выключал канал\",\nerror_severity: 3,\nerror_type: \"semantic\",\nerror_categories: \"mistranslation\"\n\nerror_comment: \"Слово «gut-wrenching» передает интенсивное эмоциональное страдание, тогда как «щемящее» звучит слишком мягко. Лучше: «мучительное чувство».\",\nerror_span: \"щемящее чувство\",\nerror_severity: 4,\nerror_type: \"semantic\",\nerror_categories: \"mistranslation\"\n\nerror_comment: \"Фраза «when it plays» относится к показу трансляции, а не к звучанию. Корректный перевод: «когда это показывают».\",\nerror_span: \"когда это звучит\",\nerror_severity: 4,\nerror_type: \"semantic\",\nerror_categories: \"mistranslation\"\n\nerror_comment: \"Неформальное «gotta» переведено формальным «мне нужно было». Следует сохранить разговорный стиль: «надо отвлечься».\",\nerror_span: \"мне нужно было отвлечься\",\nerror_severity: 2,\nerror_type: \"non-semantic\",\nerror_categories: \"style\"\n\nerror_comment: \"Слово «regularly» означает «регулярно», а не «обычно». Это искажает частоту действий.\",\nerror_span: \"обычно\",\nerror_severity: 3,\nerror_type: \"semantic\",\nerror_categories: \"mistranslation\"\n\nerror_comment: \"Тире перед «я» должно быть оформлено как длинное тире с пробелами: «глубины души — я».\",\nerror_span: \"-\",\nerror_severity: 2,\nerror_type: \"non-semantic\",\nerror_categories: \"grammar\"\n\n### Error report #3\nerror_comment: В оригинале нет кавычек, и нет указания на то, что это название программы. Скорее всего, это описание содержания трансляции.\nerror_span: \"День, когда мир был разрушен\"\nerror_severity: 3\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: В оригинале нет указания на пол говорящего, а в русском переводе используется женский род. Без контекста невозможно определить правильность выбора рода.\nerror_span: поднялась\nerror_severity: 3\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: В оригинале нет указания на пол говорящего, а в русском переводе используется женский род. Без контекста невозможно определить правильность выбора рода.\nerror_span: старалась\nerror_severity: 3\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: В оригинале нет указания на пол говорящего, а в русском переводе используется женский род. Без контекста невозможно определить правильность выбора рода.\nerror_span: устала\nerror_severity: 3\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: Неестественное для русского языка выражение. Лучше: \"вызвало во мне\".\nerror_span: вызвало у меня\nerror_severity: 2\nerror_type: non-semantic\nerror_categories: Fluency\n\nerror_comment: Метафора \"открывает шлюзы\" звучит неестествен��о в русском языке. Лучше: \"пробуждает\", \"вызывает поток\".\nerror_span: открывает шлюзы\nerror_severity: 3\nerror_type: non-semantic\nerror_categories: Fluency\n\nerror_comment: В оригинале \"granted\" используется в значении \"хотя\", \"правда\". \"Конечно\" в данном контексте меняет смысл предложения.\nerror_span: конечно\nerror_severity: 3\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: Разговорное выражение \"gotta get my mind off things\" переведено слишком формально. Лучше использовать более разговорный вариант: \"надо отвлечься\", \"надо думать о другом\".\nerror_span: мне нужно было отвлечься\nerror_severity: 2\nerror_type: non-semantic\nerror_categories: Style\n\nerror_comment: Точка с запятой используется неправильно. В русском языке она ставится между независимыми предложениями, связанными по смыслу. Здесь лучше использовать тире или двоеточие.\nerror_span: ;\nerror_severity: 2\nerror_type: non-semantic\nerror_categories: Grammar\n\n### Error Report #4\nerror_comment: \"В оригинале местоимение 'I' используется в мужском роде, что подтверждается контекстом ('Dad always found himself turning the channel off when I was unable to tear my eyes away'). Однако в переводе используется женский род ('поднялась', 'я старалась', 'я устала'), что искажает смысл текста и изменяет пол рассказчика.\",\nerror_span: \"поднялась на ноги, мне нужно было отвлечься. Его грустное настроение сохранялось, и обычно я старалась поднять ему настроение и хотя бы немного порадовать, но сегодня я устала.\",\nerror_severity: 5,\nerror_type: \"semantic\",\nerror_categories: \"Mistranslation\"\n\nerror_comment: \"Слово 'выпуска' не передает точного значения 'stream' в данном контексте, где речь идет о трансляции или потоке. Лучше использовать 'трансляция' или 'поток'.\",\nerror_span: \"выпуска\",\nerror_severity: 4,\nerror_type: \"semantic\",\nerror_categories: \"Mistranslation\"\n\nerror_comment: \"Фраза 'День, когда мир был разрушен' звучит слишком буквально и неестественно для русского языка. Лучше: 'День, когда мир рухнул' или 'День, когда мир погиб'.\",\nerror_span: \"День, когда мир был разрушен\",\nerror_severity: 3,\nerror_type: \"non-semantic\",\nerror_categories: \"Fluency\"\n\nerror_comment: \"Слово 'звучит' не соответствует контексту оригинала, где 'it plays' относится к воспроизведению видео или трансляции. Лучше: 'идет' или 'показывают'.\",\nerror_span: \"звучит\",\nerror_severity: 4,\nerror_type: \"semantic\",\nerror_categories: \"Mistranslation\"\n\nerror_comment: \"Фраза 'щемящее чувство' звучит излишне эмоционально и не соответствует нейтральному тону оригинала. Лучше: 'тяжелое чувство' или 'гнетущее чувство'.\",\nerror_span: \"щемящее чувство\",\nerror_severity: 3,\nerror_type: \"non-semantic\",\nerror_categories: \"Fluency\"\n\nerror_comment: \"Фраза 'открывает шлюзы для стольких воспоминаний' звучит неестественно для русского языка. Лучше: 'вызывает поток воспоминаний' или 'пробуждает множество воспоминаний'.\",\nerror_span: \"открывает шлюзы для стольких воспоминаний\",\nerror_severity: 3,\nerror_type: \"non-semantic\",\nerror_categories: \"Fluency\"\n\nerror_comment: \"В оригинале используется 'granted they weren't very good ones', что подразумевает уточнение, что воспоминания были плохими. Перевод 'конечно, они были не очень хорошими' звучит неестественно. Лучше: 'правда, они были далеко не самыми приятными'.\",\nerror_span: \"конечно, они были не очень хорошими\",\nerror_severity: 3,\nerror_type: \"non-semantic\",\nerror_categories: \"Fluency\"\n\nerror_comment: \"Фраза 'мне нужно было отвлечься' звучит как завершенное действие, тогда как в оригинале используется форма, указывающая на намерение ('gotta get my mind off things'). Лучше: 'нужно отвлечься'.\",\nerror_span: \"мне нужно было отвлечься\",\nerror_severity: 3,\nerror_type: \"semantic\",\nerror_categories: \"Mistranslation\"\n"
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"role": "assistant",
|
| 21 |
+
"content": "## Reports Review\nВсе отчёты, кроме второго, правильно выделяют the day the world was ruined, причём в первом отчёте дано два объяснения и верное только последнее. Третий отчёт лаконичнее всего описывает ошибку с кавычками, а четвёртый – ошибку fluency.\nОшибка when it plays выделена в 1,2 и 4 отчётах, причём лучшие объяснения даны во втором и четвёртом.\nВ 1, 3 и 4 отчётах обратили внимание на значение слова granted. Мне нравится объяснение из четвёртого отчёта, но в категориях я согласна с первым и третьим.\nФрагмент his saddened mood remained выделен только в первом отчёте, но, на мой взгляд, ошибка определена неправильно, и здесь проблема fluency со словом \"сохранялось\", а не смысловая ошибка \"грустное\".\nВо втором и четвёртом отчёте правильно выделили слово stream, причём второй отчёт даёт более простую и ёмкую причину.\nТолько во втором отчёте выделено turning the channel off, но я бы не назвала эту ошибку смысловой, тут скорее fluency.\nСлово gotta заметили во всех отчётах, но правильно распознали ошибку только во втором и третьем. Пояснение из третьего кажется мне чуть лучше.\nДефис перед \"я\" замечен только во втором отчёте, категории присвоены правильно.\n\"Шлюзы\" упомянуты в третьем и четвёртом отчётах, категории верные.\nТочка с запятой отмечена как ошибка только в третьем отчёте, категории правильные.\n\n## Final Error Report\nerror_comment: \"Выпуск\" подразумевает одну передачу из серии. В этом контексте такой перевод искажает смысл, т. к. речь идёт о трансляции единичного события.\nerror_span: повтор того же самого выпуска\nerror_severity: 4\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: Написание фразы в кавычках искажает смысл: создаётся впечатление, что оно указывает на название передачи, а не описывает произошедшие события.\nerror_span: \"День, когда мир был разрушен\"\nerror_severity: 4\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: Фраза \"День, когда мир был разрушен\" звучит слишком буквально и неестественно для русского языка. Лучше: \"День, когда мир рухнул\".\nerror_span: день, когда мир был разрушен\nerror_severity: 2\nerror_type: non-semantic\nerror_categories: Fluency\n\nerror_comment: Нехарактерная для русского языка сочетаемость. Вариант замены: \"выключал телевизор\".\nerror_span: выключал канал\nerror_severity: 3\nerror_type: non-semantic\nerror_categories: Fluency\n\nerror_comment: После \"из глубины души\" вместо дефиса должно стоять тире. Дефис используется в сложных словах и для присоединения частиц.\nerror_span: -\nerror_severity: 2\nerror_type: non-semantic\nerror_categories: grammar\n\nerror_comment: Речь идёт о трансляции, поэтому можно перевести как \"идёт\" или \"показывают\".\nerror_span: звучит\nerror_severity: 4\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: Технический термин \"шлюзы\" плохо сочетается с художественным текстом. Вариант замены: \"воспоминания накрывают меня потоком\".\nerror_span: открывает шлюзы для стольких воспоминаний\nerror_severity: 3\nerror_type: non-semantic\nerror_categories: Fluency\n\nerror_comment: Точка с запятой используется неправильно. В русском языке она ставится между независимыми предложениями, связанными по смыслу. Здесь лучше использовать тире или запятую.\nerror_span: воспоминаний;\nerror_severity: 2\nerror_type: non-semantic\nerror_categories: Grammar\n\nerror_comment: В оригинале используется \"granted they weren’t very good ones\", что подразумевает уточнение, что воспоминания были плохими. Перевод \"конечно, они были не очень хорошими\" звучит неестественно. Лучше: \"правда, они были далеко не самыми приятными\".\nerror_span: конечно\nerror_severity: 3\nerror_type: semantic\nerror_categories: Mistranslation\n\nerror_comment: Разговорное выражение \"gotta get my mind off things\" переведено слишком формально. Лучше использовать более разговорный вариант: \"надо отвлечься\", \"надо думать о другом\".\nerror_span: мне нужно было отвлечься\nerror_severity: 1\nerror_type: non-semantic\nerror_categories: Style\n\nerror_comment: Фраза \"Его грустное настроение сохранялось\" звучит неестественно, вместо \"сохранялось\" можно использовать \"не покидало его\" или \"он оставался грустным\".\nerror_span: его грустное настроение сохранялось\nerror_severity: 3\nerror_type: non-semantic\nerror_categories: Fluency\n"
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"role": "user",
|
| 25 |
+
"content": "## Source\n{{source}}\n\n## Translation\n{{translation}}\n\n## Fusing task input error reports:\n{{reports}}\n"
|
| 26 |
+
}
|
| 27 |
+
]
|
| 28 |
+
}
|
prompts/orbit_mc_fuser.md
ADDED
|
@@ -0,0 +1,477 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# ORBIT-MC fuser
|
| 2 |
+
|
| 3 |
+
## 1. `system`
|
| 4 |
+
|
| 5 |
+
````text
|
| 6 |
+
# Translation Error Reports Fusion Task
|
| 7 |
+
## Input Description
|
| 8 |
+
Several translation error reports from multiple experts are provided as input. Each report contains a comment, error type, severity level, and category. Different experts may identify different valid errors but may also include unnecessary ones. Experts may disagree on the severity level or error type for certain errors.
|
| 9 |
+
|
| 10 |
+
## Task Requirements
|
| 11 |
+
You need to combine multiple reports into one comprehensive report. Write a chain of reasoning about which errors are identified by different experts, selecting only the correct ones. Choose the most accurate comment and make a decision on the severity level and category. Essentially, you should go through the text and examine errors found by different reports, ignoring irrelevant errors.
|
| 12 |
+
|
| 13 |
+
It's important that the final report only includes errors that appear in at least one of the original reports. The fusion process should aggregate and slightly edit existing content without adding anything new.
|
| 14 |
+
|
| 15 |
+
## Expert Guidelines for Original Reports
|
| 16 |
+
Experts follow these guidelines when analyzing translations and writing error reports:
|
| 17 |
+
|
| 18 |
+
### Error Severity Scale (1 to 5):
|
| 19 |
+
4-5 (Major/Critical Errors): Significantly distort meaning, lead to misunderstanding, or are highly inappropriate
|
| 20 |
+
2-3 (Minor errors): Negatively impact readability, fluency, style, or grammar without seriously distorting essential information
|
| 21 |
+
1 (Negligible or arguable errors): Debatable stylistic nuances or very minor mistakes with minimal impact
|
| 22 |
+
### Error Types and Categories:
|
| 23 |
+
Semantic Errors (affecting meaning):
|
| 24 |
+
- Mistranslation: Incorrect translation of meaning, ambiguity, literal translations that miss idioms' meaning
|
| 25 |
+
- Undertranslation: Too general translation with lost details or nuances
|
| 26 |
+
- Overtranslation: Unjustified specific translation of general terms
|
| 27 |
+
- Omission: Deletion of information without justification
|
| 28 |
+
Non-Semantic Errors:
|
| 29 |
+
- Grammar: Grammar, punctuation, spelling mistakes not affecting meaning
|
| 30 |
+
- Style: Stylistically awkward or unnatural wording
|
| 31 |
+
- Do Not Translate: Improper translation of items that should remain untranslated
|
| 32 |
+
- Inconsistency: Variations in translations within the same text
|
| 33 |
+
- Named Entity: Incorrect translation of proper nouns
|
| 34 |
+
- Fluency: Issues affecting readability or natural flow of sentences
|
| 35 |
+
Other:
|
| 36 |
+
- Used only when no other categories apply
|
| 37 |
+
|
| 38 |
+
## After listing all errors, conclude with the final accuracy and fluency scores on a scale from 0 to 100:
|
| 39 |
+
|
| 40 |
+
Accuracy (0-100): Measures how faithfully the translation conveys the meaning of the source text. Consider whether all information is correctly transferred, whether key concepts are accurately represented, and whether there are any semantic distortions.
|
| 41 |
+
90-100: Near perfect accuracy with minimal or no semantic errors
|
| 42 |
+
70-89: Good accuracy with few minor meaning issues
|
| 43 |
+
50-69: Moderate accuracy with some notable meaning issues
|
| 44 |
+
30-49: Poor accuracy with substantial meaning distortions
|
| 45 |
+
0-29: Very poor accuracy with critical meaning errors or significant missing content
|
| 46 |
+
|
| 47 |
+
Fluency (0-100): Measures how natural, readable, and grammatically correct the translation is in the target language, regardless of accuracy. Consider grammar, word choice, idiomatic expressions, and overall readability.
|
| 48 |
+
|
| 49 |
+
90-100: Reads like original target-language text, perfect grammar and naturalness
|
| 50 |
+
70-89: Generally fluent with minor issues in style or expression
|
| 51 |
+
50-69: Somewhat fluent but with noticeable awkwardness or unnatural phrasing
|
| 52 |
+
30-49: Poor fluency with numerous grammatical errors or unnatural constructions
|
| 53 |
+
0-29: Very poor fluency, difficult to understand, severely ungrammatical
|
| 54 |
+
|
| 55 |
+
### Example
|
| 56 |
+
```
|
| 57 |
+
## Final Error Report
|
| 58 |
+
|
| 59 |
+
error_comment: Detailed explanation of the translation mistake.
|
| 60 |
+
error_span: Exact problematic text from the translation. For omission errors, describe what is missing in brackets, e.g., "[missing: recent]"
|
| 61 |
+
error_severity: Severity rating from 1 (negligible) to 5 (critical).
|
| 62 |
+
error_type: Semantic or Non-semantic.
|
| 63 |
+
error_categories: Specific category (choose from mistranslation, undertranslation, overtranslation, omission, grammar, style, do not translate, inconsistency, NE, fluency).
|
| 64 |
+
|
| 65 |
+
## Final scores:
|
| 66 |
+
Accuracy: [Number from 0 to 100]
|
| 67 |
+
Fluency: [Number from 0 to 100]
|
| 68 |
+
```
|
| 69 |
+
````
|
| 70 |
+
|
| 71 |
+
## 2. `user`
|
| 72 |
+
|
| 73 |
+
```text
|
| 74 |
+
## Source
|
| 75 |
+
The Biden Administration Is Letting Corporate Criminals Off the Hook
|
| 76 |
+
|
| 77 |
+
## Translation
|
| 78 |
+
Администрация Байдена освобождает от ответственности корпоративных преступников
|
| 79 |
+
|
| 80 |
+
## Fusing task input error reports:
|
| 81 |
+
### Error report #1
|
| 82 |
+
error_comment: "Перевод слишком буквальный и не передает идиоматического значения выражения 'letting off the hook', которое означает 'избежать наказания' или 'оставить безнаказанным'. Более подходящий перевод: 'Администраци�� Байдена позволяет корпоративным преступникам избежать наказания'.",
|
| 83 |
+
error_span: "освобождает от ответственности",
|
| 84 |
+
error_severity: 4,
|
| 85 |
+
error_type: "semantic",
|
| 86 |
+
error_categories: "Mistranslation"
|
| 87 |
+
|
| 88 |
+
### Error report #2
|
| 89 |
+
error_comment: Идиома "let off the hook" означает "позволить избежать наказания", "дать уйти от ответственности". Перевод "освобождает от ответственности" предполагает активное действие со стороны администрации, тогда как оригинал подразумевает попустительство, бездействие. Более точный перевод: "позволяет уйти от ответственности" или "закрывает глаза на действия".
|
| 90 |
+
error_span: освобождает от ответственности
|
| 91 |
+
error_severity: 3
|
| 92 |
+
error_type: semantic
|
| 93 |
+
error_categories: Mistranslation
|
| 94 |
+
|
| 95 |
+
error_comment: В русском языке словосочетание "корпоративные преступники" звучит неестественно и может быть неправильно понято. Лучше: "преступников из корпораций" или "преступления корпораций".
|
| 96 |
+
error_span: корпоративных преступников
|
| 97 |
+
error_severity: 2
|
| 98 |
+
error_type: non-semantic
|
| 99 |
+
error_categories: Fluency
|
| 100 |
+
|
| 101 |
+
error_comment: В русском языке при упоминании действующего президента США принято использовать полное имя: «администрация Джо Байдена». Использование только фамилии может восприниматься как фамильярность.
|
| 102 |
+
error_span: Байдена
|
| 103 |
+
error_severity: 2
|
| 104 |
+
error_type: non-semantic
|
| 105 |
+
error_categories: Style
|
| 106 |
+
|
| 107 |
+
error_comment: В заголовке статьи используется настоящее длительное время (Present Continuous), указывающее на продолжающийся процесс. В переводе использован глагол совершенного вида, что меняет смысл на единичное, завершенное действие. Лучше использовать несовершенный вид: «позволяет уходить от ответственности».
|
| 108 |
+
error_span: освобождает
|
| 109 |
+
error_severity: 3
|
| 110 |
+
error_type: semantic
|
| 111 |
+
error_categories: Mistranslation
|
| 112 |
+
|
| 113 |
+
### Error report #3
|
| 114 |
+
error_comment: The phrase "letting corporate criminals off the hook" is an idiomatic expression in English, meaning to allow corporate criminals to avoid accountability or punishment. The translation "освобождает от ответственности" is overly literal and does not fully capture the idiomatic nuance of the original, which implies a more active or deliberate allowance of impunity. A better translation might be "позволяет корпоративным преступникам избежать наказания" or "спускает корпоративным преступникам с рук."
|
| 115 |
+
error_span: освобождает от ответственности
|
| 116 |
+
error_severity: 4
|
| 117 |
+
error_type: semantic
|
| 118 |
+
error_categories: Mistranslation
|
| 119 |
+
|
| 120 |
+
error_comment: The translation does not fully convey the nuance of "letting off the hook," which implies a deliberate or lenient action on the part of the Biden Administration. The phrase "освобождает от ответственности" is too neutral and lacks the implied criticism or intentionality present in the source text. A more nuanced translation could include a phrase like "намеренно позволяет избежать ответственности" to better reflect the source's tone.
|
| 121 |
+
error_span: освобождает от ответственности
|
| 122 |
+
error_severity: 3
|
| 123 |
+
error_type: semantic
|
| 124 |
+
error_categories: Undertranslation
|
| 125 |
+
|
| 126 |
+
### Error Report #4
|
| 127 |
+
**Error Comment:** Идиома "letting off the hook" (досл. "спускать с крючка") передана буквально как "освобождает от ответственности", что теряет стилистическую окраску оригинала. В русском языке устоявшийся эквивалент — "спускать с крючка" или "позволять уйти от наказания", который лучше отражает ироничный/разговорный тон заголовка.
|
| 128 |
+
**Error Span:** освобождает от ответственности
|
| 129 |
+
**Error Severity:** 3
|
| 130 |
+
**Error Type:** non-semantic
|
| 131 |
+
**Error Categories:** Fluency
|
| 132 |
+
|
| 133 |
+
---
|
| 134 |
+
|
| 135 |
+
**Error Comment:** Отсутствие кавычек или курсива для выделения идиоматического выражения (хотя это необязательно, но может быть важно для сохранения стиля).
|
| 136 |
+
**Error Span:** корпоративных преступников
|
| 137 |
+
**Error Severity:** 1 (деbatable)
|
| 138 |
+
**Error Type:** non-semantic
|
| 139 |
+
**Error Categories:** Style
|
| 140 |
+
```
|
| 141 |
+
|
| 142 |
+
## 3. `assistant`
|
| 143 |
+
|
| 144 |
+
```text
|
| 145 |
+
## Reports Review
|
| 146 |
+
Все четыре отчёта верно выделяют фрагмент "letting off the hook", но только в отчётах 2 и 3 даётся верное объяснение ошибки: в оригинале подразумевается пассивность властей, а не активные действия по освобождению корпораций от ответственности.
|
| 147 |
+
|
| 148 |
+
Только отчёт 2 верно выделяет ошибку в сочетании "корпоративные преступники", однако легко заметить противоречие между комментарием, где есть слова "словосочетание может быть неправильно понято", и категорией и типом ошибки (non-semantic, Fluency): раз перевод вводит читателя в заблуждение, тип должен определяться как semantic, Mistranslation. Поскольку в целом искажение не слишком критичное, его степень можно оценить на 3.
|
| 149 |
+
|
| 150 |
+
## Final Error Report
|
| 151 |
+
error_comment: Идиома "let off the hook" означает "позволить избежать наказания", "дать уйти от ответственности". Перевод "освобождает от ответственности" предполагает активное действие со стороны администрации, тогда как оригинал подразумевает попустительство, бездействие. Более точный перевод: "позволяет уйти от ответственности" или "закрывает глаза на действия".
|
| 152 |
+
error_span: освобождает от ответственности
|
| 153 |
+
error_severity: 3
|
| 154 |
+
error_type: semantic
|
| 155 |
+
error_categories: Mistranslation
|
| 156 |
+
|
| 157 |
+
error_comment: В оригинальном тексте corporate crime относится к преступлениям, которые совершают либо корпорации, либо отдельные люди действующие от лица корпорации. Перевод "корпоративные преступники" является смысловой неточностью, так как в русском слово "преступники" относится только к людям, тогда как в оригинальном тексте под corporate criminals подразумеваются корпорации. Правильнее было бы перевести "корпоративные преступления" или "преступления корпораций".
|
| 158 |
+
error_span: корпоративных преступников
|
| 159 |
+
error_severity: 3
|
| 160 |
+
error_type: semantic
|
| 161 |
+
error_categories: Mistranslation
|
| 162 |
+
```
|
| 163 |
+
|
| 164 |
+
## 4. `user`
|
| 165 |
+
|
| 166 |
+
```text
|
| 167 |
+
## Source
|
| 168 |
+
A click. It had been reruns of this same stream, the day the world was ruined. Dad always found himself turning the channel off when I was unable to tear my eyes away. The somber face he gave me ensured a gut-wrenching feeling transpiring from deep within – I always feel bad when it plays, but it opens the floodgates to so many memories; granted they weren't very good ones. Letting out a sigh, I slowly rose to my feet, gotta get my mind off things. His saddened mood remained, and regularly I’d try to help ease his mood and at least give him a little bit of joy, but today; I’m tired.
|
| 169 |
+
|
| 170 |
+
## Translation
|
| 171 |
+
Щелчок. Это был повтор того же самого выпуска "День, когда мир был разрушен". Папа всегда выключал канал, когда я не мог оторвать от него глаз. Мрачное выражение его лица вызвало у меня щемящее чувство, поднимающееся из глубины души - я всегда чувствую себя плохо, когда это звучит, но это открывает шлюзы для стольких воспоминаний; конечно, они были не очень хорошими. Вздохнув, я медленно поднялась на ноги, мне нужно было отвлечься. Его грустное настроение сохранялось, и обычно я старалась поднять ему настроение и хотя бы немного порадовать, но сегодня я устала.
|
| 172 |
+
|
| 173 |
+
## Fusing task input error reports:
|
| 174 |
+
### Error report #1
|
| 175 |
+
error_comment: The phrase "the day the world was ruined" is mistranslated as "День, когда мир был разрушен," which implies a specific event or day. However, in the source, it refers to the title of the stream or broadcast. A more accurate translation would be "выпуск под названием 'День, когда мир был разрушен'" to clarify that it is a title and not a literal description.
|
| 176 |
+
error_span: "День, когда мир был разрушен"
|
| 177 |
+
error_severity: 4
|
| 178 |
+
error_type: semantic
|
| 179 |
+
error_categories: Mistranslation
|
| 180 |
+
|
| 181 |
+
error_comment: The phrase "gut-wrenching feeling transpiring from deep within" is mistranslated as "щемящее чувство, поднимающееся из глубины души." While "щемящее чувство" captures part of the meaning, "поднимающееся из глубины души" shifts the imagery and emotional intensity. A better translation would be "щемящее чувство, исходящее из глубины души," which maintains the original tone and imagery.
|
| 182 |
+
error_span: "щемящее чувство, поднимающееся из глубины души"
|
| 183 |
+
error_severity: 3
|
| 184 |
+
error_type: semantic
|
| 185 |
+
error_categories: Mistranslation
|
| 186 |
+
|
| 187 |
+
error_comment: The phrase "I always feel bad when it plays" is mistranslated as "я всегда чувствую себя плохо, когда это звучит." The word "звучит" (sounds) is incorrect because it implies audio-only content, whereas the original refers to a video stream. A better translation would be "я всегда чувствую себя плохо, когда это показывают."
|
| 188 |
+
error_span: "когда это звучит"
|
| 189 |
+
error_severity: 4
|
| 190 |
+
error_type: semantic
|
| 191 |
+
error_categories: Mistranslation
|
| 192 |
+
|
| 193 |
+
error_comment: The phrase "granted they weren't very good ones" is mistranslated as "конечно, они были не очень хорошими." The word "конечно" (of course) adds an unwarranted assumption that the memories being bad is obvious, which is not present in the original. A better translation would be "правда, они были не очень хорошими."
|
| 194 |
+
error_span: "конечно, они были не очень хорошими"
|
| 195 |
+
error_severity: 3
|
| 196 |
+
error_type: semantic
|
| 197 |
+
error_categories: Mistranslation
|
| 198 |
+
|
| 199 |
+
error_comment: The phrase "gotta get my mind off things" is mistranslated as "мне нужно было отвлечься." The past tense "нужно было" (needed to) changes the meaning, as the original is in the present tense, expressing an immediate need. A better translation would be "мне нужно отвлечься."
|
| 200 |
+
error_span: "мне нужно было отвлечься"
|
| 201 |
+
error_severity: 4
|
| 202 |
+
error_type: semantic
|
| 203 |
+
error_categories: Mistranslation
|
| 204 |
+
|
| 205 |
+
error_comment: The phrase "His saddened mood remained" is mistranslated as "Его грустное настроение сохранялось." While technically correct, the translation loses the emotional nuance of "saddened mood," which conveys a deeper, more personal sadness. A better translation would be "Его печальное настроение сохранялось."
|
| 206 |
+
error_span: "Его грустное настроение сохранялось"
|
| 207 |
+
error_severity: 3
|
| 208 |
+
error_type: semantic
|
| 209 |
+
error_categories: Mistranslation
|
| 210 |
+
|
| 211 |
+
error_comment: The phrase "regularly I’d try to help ease his mood and at least give him a little bit of joy" is mistranslated as "обычно я старалась поднять ему настроение и хотя бы немного порадовать." The word "обычно" (usually) is less frequent than "regularly" and slightly shifts the meaning. A better translation would be "я регулярно старалась поднять ему настроение и хотя бы немного порадовать."
|
| 212 |
+
error_span: "обычно я старалась"
|
| 213 |
+
error_severity: 3
|
| 214 |
+
error_type: semantic
|
| 215 |
+
error_categories: Mistranslation
|
| 216 |
+
|
| 217 |
+
error_comment: The phrase "but today; I’m tired" is mistranslated as "но сегодня я устала." The use of "устала" (feminine form) assumes the narrator is female, while the original text does not specify the narrator's gender. A gender-neutral translation would be "но сегодня я устал(а)."
|
| 218 |
+
error_span: "но сегодня я устала"
|
| 219 |
+
error_severity: 4
|
| 220 |
+
error_type: semantic
|
| 221 |
+
error_categories: Mistranslation
|
| 222 |
+
|
| 223 |
+
error_comment: The phrase "A click" is undertranslated as "Щелчок," which omits the implied action of turning on or off a device. A more complete translation would be "Щелчок переключения" or "Щелчок, раздавшийся при переключении канала."
|
| 224 |
+
error_span: "Щелчок"
|
| 225 |
+
error_severity: 3
|
| 226 |
+
error_type: semantic
|
| 227 |
+
error_categories: Undertranslation
|
| 228 |
+
|
| 229 |
+
error_comment: The phrase "the day the world was ruined" in the original text implies a specific historical or fictional event tied to the stream, but this nuance is not fully conveyed in the translation. The omission of this context makes the title less impactful. Adding "о дне, когда мир был разрушен" would better capture the original meaning.
|
| 230 |
+
error_span: "День, когда мир был разрушен"
|
| 231 |
+
error_severity: 3
|
| 232 |
+
error_type: semantic
|
| 233 |
+
error_categories: Omission
|
| 234 |
+
|
| 235 |
+
### Error report #2
|
| 236 |
+
error_comment: "Слово «stream» в данном контексте относится к повторяющейся трансляции, а не к выпуску. Более точный перевод — «повтор той же трансляции».",
|
| 237 |
+
error_span: "повтор того же самого выпуска",
|
| 238 |
+
error_severity: 3,
|
| 239 |
+
error_type: "semantic",
|
| 240 |
+
error_categories: "mistranslation"
|
| 241 |
+
|
| 242 |
+
error_comment: "Фраза «turning the channel off» означает «переключить канал», а не «выключить канал». В русском языке корректнее: «переключал канал» или «выключал телевизор».",
|
| 243 |
+
error_span: "выключал канал",
|
| 244 |
+
error_severity: 3,
|
| 245 |
+
error_type: "semantic",
|
| 246 |
+
error_categories: "mistranslation"
|
| 247 |
+
|
| 248 |
+
error_comment: "Слово «gut-wrenching» передает интенсивное эмоциональное страдание, тогда как «щемящее» звучит слишком мягко. Лучше: «мучительное чувство».",
|
| 249 |
+
error_span: "щемящее чувство",
|
| 250 |
+
error_severity: 4,
|
| 251 |
+
error_type: "semantic",
|
| 252 |
+
error_categories: "mistranslation"
|
| 253 |
+
|
| 254 |
+
error_comment: "Фраза «when it plays» относится к показу трансляции, а не к звучанию. Корректный перевод: «когда это показывают».",
|
| 255 |
+
error_span: "когда это звучит",
|
| 256 |
+
error_severity: 4,
|
| 257 |
+
error_type: "semantic",
|
| 258 |
+
error_categories: "mistranslation"
|
| 259 |
+
|
| 260 |
+
error_comment: "Неформальное «gotta» переведено формальным «мне нужно было». Следует сохранить разговорный стиль: «надо отвлечься».",
|
| 261 |
+
error_span: "мне нужно было отвлечься",
|
| 262 |
+
error_severity: 2,
|
| 263 |
+
error_type: "non-semantic",
|
| 264 |
+
error_categories: "style"
|
| 265 |
+
|
| 266 |
+
error_comment: "Слово «regularly» означает «регулярно», а не «обычно». Это искажает частоту действий.",
|
| 267 |
+
error_span: "обычно",
|
| 268 |
+
error_severity: 3,
|
| 269 |
+
error_type: "semantic",
|
| 270 |
+
error_categories: "mistranslation"
|
| 271 |
+
|
| 272 |
+
error_comment: "Тире перед «я» должно быть оформлено как длинное тире с пробелами: «глубины души — я».",
|
| 273 |
+
error_span: "-",
|
| 274 |
+
error_severity: 2,
|
| 275 |
+
error_type: "non-semantic",
|
| 276 |
+
error_categories: "grammar"
|
| 277 |
+
|
| 278 |
+
### Error report #3
|
| 279 |
+
error_comment: В оригинале нет кавычек, и нет указания на то, что это название программы. Скорее всего, это описание содержания трансляции.
|
| 280 |
+
error_span: "День, когда мир был разрушен"
|
| 281 |
+
error_severity: 3
|
| 282 |
+
error_type: semantic
|
| 283 |
+
error_categories: Mistranslation
|
| 284 |
+
|
| 285 |
+
error_comment: В оригинале нет указания на пол говорящего, а в русском переводе используется женский род. Без контекста невозможно определить правильность выбора рода.
|
| 286 |
+
error_span: поднялась
|
| 287 |
+
error_severity: 3
|
| 288 |
+
error_type: semantic
|
| 289 |
+
error_categories: Mistranslation
|
| 290 |
+
|
| 291 |
+
error_comment: В оригинале нет указания на пол говорящего, а в русском переводе используется женский род. Без контекста невозможно определить правильность выбора рода.
|
| 292 |
+
error_span: старалась
|
| 293 |
+
error_severity: 3
|
| 294 |
+
error_type: semantic
|
| 295 |
+
error_categories: Mistranslation
|
| 296 |
+
|
| 297 |
+
error_comment: В оригинале нет указания на пол говорящего, а в русском переводе используется женский род. Без контекста невозможно определить правильность выбора рода.
|
| 298 |
+
error_span: устала
|
| 299 |
+
error_severity: 3
|
| 300 |
+
error_type: semantic
|
| 301 |
+
error_categories: Mistranslation
|
| 302 |
+
|
| 303 |
+
error_comment: Неестественное для русского языка выражение. Лучше: "вызвало во мне".
|
| 304 |
+
error_span: вызвало у меня
|
| 305 |
+
error_severity: 2
|
| 306 |
+
error_type: non-semantic
|
| 307 |
+
error_categories: Fluency
|
| 308 |
+
|
| 309 |
+
error_comment: Метафора "открывает шлюзы" звучит неестественно в русском языке. Лучше: "пробуждает", "вызывает поток".
|
| 310 |
+
error_span: открывает шлюзы
|
| 311 |
+
error_severity: 3
|
| 312 |
+
error_type: non-semantic
|
| 313 |
+
error_categories: Fluency
|
| 314 |
+
|
| 315 |
+
error_comment: В оригинале "granted" используется в значении "хотя", "правда". "��онечно" в данном контексте меняет смысл предложения.
|
| 316 |
+
error_span: конечно
|
| 317 |
+
error_severity: 3
|
| 318 |
+
error_type: semantic
|
| 319 |
+
error_categories: Mistranslation
|
| 320 |
+
|
| 321 |
+
error_comment: Разговорное выражение "gotta get my mind off things" переведено слишком формально. Лучше использовать более разговорный вариант: "надо отвлечься", "надо думать о другом".
|
| 322 |
+
error_span: мне нужно было отвлечься
|
| 323 |
+
error_severity: 2
|
| 324 |
+
error_type: non-semantic
|
| 325 |
+
error_categories: Style
|
| 326 |
+
|
| 327 |
+
error_comment: Точка с запятой используется неправильно. В русском языке она ставится между независимыми предложениями, связанными по смыслу. Здесь лучше использовать тире или двоеточие.
|
| 328 |
+
error_span: ;
|
| 329 |
+
error_severity: 2
|
| 330 |
+
error_type: non-semantic
|
| 331 |
+
error_categories: Grammar
|
| 332 |
+
|
| 333 |
+
### Error Report #4
|
| 334 |
+
error_comment: "В оригинале местоимение 'I' используется в мужском роде, что подтверждается контекстом ('Dad always found himself turning the channel off when I was unable to tear my eyes away'). Однако в переводе используется женский род ('поднялась', 'я старалась', 'я устала'), что искажает смысл текста и изменяет пол рассказчика.",
|
| 335 |
+
error_span: "поднялась на ноги, мне нужно было отвлечься. Его грустное настроение сохранялось, и обычно я старалась поднять ему настроение и хотя бы немного порадовать, но сегодня я устала.",
|
| 336 |
+
error_severity: 5,
|
| 337 |
+
error_type: "semantic",
|
| 338 |
+
error_categories: "Mistranslation"
|
| 339 |
+
|
| 340 |
+
error_comment: "Слово 'выпуска' не передает точного значения 'stream' в данном контексте, где речь идет о трансляции или потоке. Лучше использовать 'трансляция' или 'поток'.",
|
| 341 |
+
error_span: "выпуска",
|
| 342 |
+
error_severity: 4,
|
| 343 |
+
error_type: "semantic",
|
| 344 |
+
error_categories: "Mistranslation"
|
| 345 |
+
|
| 346 |
+
error_comment: "Фраза 'День, когда мир был разрушен' звучит слишком буквально и неестественно для русского языка. Лучше: 'День, когда мир рухнул' или 'День, когда мир погиб'.",
|
| 347 |
+
error_span: "День, когда мир был разрушен",
|
| 348 |
+
error_severity: 3,
|
| 349 |
+
error_type: "non-semantic",
|
| 350 |
+
error_categories: "Fluency"
|
| 351 |
+
|
| 352 |
+
error_comment: "Слово 'звучит' не соответствует контексту оригинала, где 'it plays' относится к воспроизведению видео или трансляции. Лучше: 'идет' или 'показывают'.",
|
| 353 |
+
error_span: "звучит",
|
| 354 |
+
error_severity: 4,
|
| 355 |
+
error_type: "semantic",
|
| 356 |
+
error_categories: "Mistranslation"
|
| 357 |
+
|
| 358 |
+
error_comment: "Фраза 'щемящее чувство' звучит излишне эмоционально и не соответствует нейтральному тону оригинала. Лучше: 'тяжелое чувство' или 'гнетущее чувство'.",
|
| 359 |
+
error_span: "щемящее чувство",
|
| 360 |
+
error_severity: 3,
|
| 361 |
+
error_type: "non-semantic",
|
| 362 |
+
error_categories: "Fluency"
|
| 363 |
+
|
| 364 |
+
error_comment: "Фраза 'открывает шлюзы для стольких воспоминаний' звучит неестественно для русского языка. Лучше: 'вызывает поток воспоминаний' или 'пробуждает множество воспоминаний'.",
|
| 365 |
+
error_span: "открывает шлюзы для стольких воспоминаний",
|
| 366 |
+
error_severity: 3,
|
| 367 |
+
error_type: "non-semantic",
|
| 368 |
+
error_categories: "Fluency"
|
| 369 |
+
|
| 370 |
+
error_comment: "В оригинале используется 'granted they weren't very good ones', что подразумевает уточнение, что воспоминания были плохими. Перевод 'конечно, они были не очень хорошими' звучит неестественно. Лучше: 'правда, они были далеко не самыми приятными'.",
|
| 371 |
+
error_span: "конечно, они были не очень хорошими",
|
| 372 |
+
error_severity: 3,
|
| 373 |
+
error_type: "non-semantic",
|
| 374 |
+
error_categories: "Fluency"
|
| 375 |
+
|
| 376 |
+
error_comment: "Фраза 'мне нужно было отвлечься' звучит как завершенное действие, тогда как в оригинале используется форма, указывающая на намерение ('gotta get my mind off things'). Лучше: 'нужно отвлечься'.",
|
| 377 |
+
error_span: "мне нужно было отвлечься",
|
| 378 |
+
error_severity: 3,
|
| 379 |
+
error_type: "semantic",
|
| 380 |
+
error_categories: "Mistranslation"
|
| 381 |
+
```
|
| 382 |
+
|
| 383 |
+
## 5. `assistant`
|
| 384 |
+
|
| 385 |
+
```text
|
| 386 |
+
## Reports Review
|
| 387 |
+
Все отчёты, кроме второго, правильно выделяют the day the world was ruined, причём в первом отчёте дано два объяснения и верное только последнее. Третий отчёт лаконичнее всего описывает ошибку с кавычками, а четвёртый – ошибку fluency.
|
| 388 |
+
Ошибка when it plays выделена в 1,2 и 4 отчётах, причём лучшие объяснения даны во втором и четвёртом.
|
| 389 |
+
В 1, 3 и 4 отчётах обратили внимание на значение слова granted. Мне нравится объяснение из четвёртого отчёта, но в категориях я согласна с первым и третьим.
|
| 390 |
+
Фрагмент his saddened mood remained выделен только в первом отчёте, но, на мой взгляд, ошибка определена неправильно, и здесь проблема fluency со словом "сохранялось", а не смысловая ошибка "грустное".
|
| 391 |
+
Во втором и четвёртом отчёте правильно выделили слово stream, причём второй отчёт даёт более простую и ёмкую причину.
|
| 392 |
+
Только во втором отчёте выделено turning the channel off, но я бы не назвала эту ошибку смысловой, тут скорее fluency.
|
| 393 |
+
Слово gotta заметили во всех отчётах, но правильно распознали ошибку только во втором и третьем. Пояснение из третьего кажется мне чуть лучше.
|
| 394 |
+
Дефис перед "я" замечен только во втором отчёте, категории присвоены правильно.
|
| 395 |
+
"Шлюзы" упомянуты в третьем и четвёртом отчётах, категории верные.
|
| 396 |
+
Точка с запятой отмечена как ошибка только в третьем отчёте, категории правильные.
|
| 397 |
+
|
| 398 |
+
## Final Error Report
|
| 399 |
+
error_comment: "Выпуск" подразумевает одну передачу из серии. В этом контексте такой перевод искажает смысл, т. к. речь идёт о трансляции единичного события.
|
| 400 |
+
error_span: повтор того же самого выпуска
|
| 401 |
+
error_severity: 4
|
| 402 |
+
error_type: semantic
|
| 403 |
+
error_categories: Mistranslation
|
| 404 |
+
|
| 405 |
+
error_comment: Написание фразы в кавычках искажает смысл: создаётся впечатление, что оно указывает на название передачи, а не описывает произошедшие события.
|
| 406 |
+
error_span: "День, когда мир был разрушен"
|
| 407 |
+
error_severity: 4
|
| 408 |
+
error_type: semantic
|
| 409 |
+
error_categories: Mistranslation
|
| 410 |
+
|
| 411 |
+
error_comment: Фраза "День, когда мир был разрушен" звучит слишком буквально и неестественно для русского языка. Лучше: "День, когда мир рухнул".
|
| 412 |
+
error_span: день, когда мир был разрушен
|
| 413 |
+
error_severity: 2
|
| 414 |
+
error_type: non-semantic
|
| 415 |
+
error_categories: Fluency
|
| 416 |
+
|
| 417 |
+
error_comment: Нехарактерная для русского языка сочетаемость. Вариант замены: "выключал телевизор".
|
| 418 |
+
error_span: выключал канал
|
| 419 |
+
error_severity: 3
|
| 420 |
+
error_type: non-semantic
|
| 421 |
+
error_categories: Fluency
|
| 422 |
+
|
| 423 |
+
error_comment: После "из глубины души" вместо дефиса должно стоять тире. Дефис используется в сложных словах и для присоединения частиц.
|
| 424 |
+
error_span: -
|
| 425 |
+
error_severity: 2
|
| 426 |
+
error_type: non-semantic
|
| 427 |
+
error_categories: grammar
|
| 428 |
+
|
| 429 |
+
error_comment: Речь идёт о трансляции, поэтому можно перевести как "идёт" или "показывают".
|
| 430 |
+
error_span: звучит
|
| 431 |
+
error_severity: 4
|
| 432 |
+
error_type: semantic
|
| 433 |
+
error_categories: Mistranslation
|
| 434 |
+
|
| 435 |
+
error_comment: Технический термин "шлюзы" плохо сочетается с художественным текстом. Вариант замены: "воспоминания накрывают меня потоком".
|
| 436 |
+
error_span: открывает шлюзы для стольких воспоминаний
|
| 437 |
+
error_severity: 3
|
| 438 |
+
error_type: non-semantic
|
| 439 |
+
error_categories: Fluency
|
| 440 |
+
|
| 441 |
+
error_comment: Точка с запятой используется неправильно. В русском я��ыке она ставится между независимыми предложениями, связанными по смыслу. Здесь лучше использовать тире или запятую.
|
| 442 |
+
error_span: воспоминаний;
|
| 443 |
+
error_severity: 2
|
| 444 |
+
error_type: non-semantic
|
| 445 |
+
error_categories: Grammar
|
| 446 |
+
|
| 447 |
+
error_comment: В оригинале используется "granted they weren’t very good ones", что подразумевает уточнение, что воспоминания были плохими. Перевод "конечно, они были не очень хорошими" звучит неестественно. Лучше: "правда, они были далеко не самыми приятными".
|
| 448 |
+
error_span: конечно
|
| 449 |
+
error_severity: 3
|
| 450 |
+
error_type: semantic
|
| 451 |
+
error_categories: Mistranslation
|
| 452 |
+
|
| 453 |
+
error_comment: Разговорное выражение "gotta get my mind off things" переведено слишком формально. Лучше использовать более разговорный вариант: "надо отвлечься", "надо думать о другом".
|
| 454 |
+
error_span: мне нужно было отвлечься
|
| 455 |
+
error_severity: 1
|
| 456 |
+
error_type: non-semantic
|
| 457 |
+
error_categories: Style
|
| 458 |
+
|
| 459 |
+
error_comment: Фраза "Его грустное настроение сохранялось" звучит неестественно, вместо "сохранялось" можно использовать "не покидало его" или "он оставался грустным".
|
| 460 |
+
error_span: его грустное настроение сохранялось
|
| 461 |
+
error_severity: 3
|
| 462 |
+
error_type: non-semantic
|
| 463 |
+
error_categories: Fluency
|
| 464 |
+
```
|
| 465 |
+
|
| 466 |
+
## 6. `user`
|
| 467 |
+
|
| 468 |
+
```text
|
| 469 |
+
## Source
|
| 470 |
+
{{source}}
|
| 471 |
+
|
| 472 |
+
## Translation
|
| 473 |
+
{{translation}}
|
| 474 |
+
|
| 475 |
+
## Fusing task input error reports:
|
| 476 |
+
{{reports}}
|
| 477 |
+
```
|
prompts/orbit_sc.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"messages": [
|
| 3 |
+
{
|
| 4 |
+
"role": "system",
|
| 5 |
+
"content": "# Translation Error Analysis Task\n\n## Instructions\nYou are an expert translator. Analyze the provided source text and its translation. Identify any errors and provide overall quality scores.\n\n### Error Severity Scale (1 to 5):\n- 4-5 (Major/Critical Errors)\n- 2-3 (Minor errors)\n- 1 (Negligible or arguable errors)\n## Error Types and Categories\n\n**Semantic errors** (affect meaning):\n- mistranslation: incorrect meaning or terminology\n- undertranslation: too general, lost nuances\n- overtranslation: added unjustified specifics\n- omission: deleted information\n\n**Non-semantic errors** (don't change meaning):\n- grammar: grammar, punctuation, spelling errors\n- style: awkward or unnatural wording\n- do not translate: incorrectly translated code, URLs, etc.\n- inconsistency: inconsistent terminology within text\n- ne: incorrect proper noun translation/transcription\n- fluency: unnatural flow, confusing structure\n- other: errors not fitting above categories\n\n## Final Scores (0-100):\n- **Accuracy**: How faithfully meaning is conveyed\n- **Fluency**: How natural and readable the translation is\n\n## Output Format\nFor each error, provide:\n```\nerror_comment: [explanation in target language]\nerror_span: [problematic text from translation] For omission errors, describe what is missing in brackets, e.g., \"[missing: recent]\"\nerror_severity: [1-5]\nerror_type: [semantic/non-semantic]\nerror_categories: [category name]\n```\n\nSeparate errors with two line breaks. End with:\n```\n## Final scores:\nAccuracy: [0-100]\nFluency: [0-100]\n```\n\nIf no errors found, output: \"No translation mistakes found.\" then Final scores.\n"
|
| 6 |
+
},
|
| 7 |
+
{
|
| 8 |
+
"role": "user",
|
| 9 |
+
"content": "## Text for Analysis\n### Source Text\nOn 8 September 2019, Bugatti presented a limited production high performance variant of the Chiron called the Chiron Super Sport 300+ to owners taking part in the company's 110th-anniversary tour in Europe. The variant is limited to 30 examples at a net unit price of 3.5 million euros. Changes from the standard Chiron include a more powerful engine, increased top speed, and a characteristic paint scheme featuring a lacquer-coated carbon fibre body with orange stripes, echoing the aesthetics of the company's previous top speed record-breaking cars, the Veyron Super Sport World Record Edition and the Veyron Grand Sport Vitesse World Record Car.\n### Translation\n8 сентября 2019 года Bugatti представила ограниченную серию высокопроизводительной модификации Chiron под названием Chiron Super Sport 300+ владельцам, участвовавшим в европейском туре компании по случаю её 110-летия. Модификация выпущена тиражом в 30 экземпляров по чистой цене 3,5 миллиона евро за единицу. Отличия от стандартного Chiron включают более мощный двигатель, увеличенную максимальную скорость и характерную окраску: лакокрасочное покрытие кузова из углепластика с оранжевыми полосами, что перекликается с эстетикой предыдущих рекордных автомобилей компании — Veyron Super Sport World Record Edition и Veyron Grand Sport Vitesse World Record Car.\n"
|
| 10 |
+
},
|
| 11 |
+
{
|
| 12 |
+
"role": "assistant",
|
| 13 |
+
"content": "## Error Report\nerror_comment: Нарушена лексическая сочетаемость, стоит заменить на «лимитированную серию» или обыграть «ограниченный выпуск»\nerror_span: ограниченную серию\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Необходима конкретизация: владельцы чего именно имеются в виду («автомобилей / гиперкаров Bugatti»)\nerror_span: владельцам\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Опущена информация о том, что эти автомобили побили рекорды скорости + неестественная конструкция; стоит переписать фрагмент, например, «выпущенных Bugatti автомобилей, побивших рекорды скорости»\nerror_span: рекордных\nerror_severity: 5\nerror_type: Semantic\nerror_categories: undertranslation\n\nerror_comment: Дословный перевод, стоит заменить на «более мощной» или «высокоскоростной»\nerror_span: высокопроизводительной\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Неверный контекстуальный синоним: «тираж» относится к печатной продукции\nerror_span: тиражом\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Неясная отнесённость\nerror_span: что\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Дословный перевод герундия, стоит заменить на «которые принимали / приняли участие»\nerror_span: участвовавшим\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Дословный перевод, нарушающий сочетаемость и создающий двусмысленность («включать двигатель» в значении его запуска)\nerror_span: включают\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Неверный контекстуальный синоним: «единица» создаёт ассоциации с научным текстом, стоит заменить на «каждый»\nerror_span: единицу\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Конструкция переусложнена: лучше «От стандартного Chiron модель отличается более мощным двигателем...»\nerror_span: Отличия от стандартного Chiron\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Громоздкий дословный перевод; стоит разбить на смысловые группы и переписать фрагмент\nerror_span: углепластика с оранжевыми полосами\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Нарушена лексическая и контекстуальная сочетаемость, лишние значения: «лакокрасочными» бывают работы, здесь речь о покрытии кузова\nerror_span: лакокрасочное покрытие\nerror_severity: 4\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Лишний оттенок значения, стоит заменить на «особую»\nerror_span: характерную\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\n## Final scores:\nAccuracy: 78\nFluency: 57\n"
|
| 14 |
+
},
|
| 15 |
+
{
|
| 16 |
+
"role": "user",
|
| 17 |
+
"content": "## Text for Analysis\n### Source Text\nYou, you the world! You're full of surprises, aren't you? I remember a few years back, watching some of those Resident Evil movies with Mila Jovovich and thinking, I can't get any worse than this. Yeah, it feels like a simpler time. But with 2021's final gasping breath that came out with Resident Evil: Welcome to Raccoon City. AKA, Welcome to Hell.\n### Translation\nТы, весь мир! Ты полон сюрпризов, не так ли? Помню, несколько лет назад я смотрел некоторые фильмы из серии «Обитель зла» с Милой Йовович и думал: «Хуже уже быть не может». Да, тогда казалось, что времена были проще. Но на последнем издыхании 2021 года вышел фильм «Обитель зла: Добро пожаловать в Раккун-Сити». Он же — «Добро пожаловать в ад».\n"
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"role": "assistant",
|
| 21 |
+
"content": "## Error Report\nerror_comment: Верное написание: Милла.\nerror_span: Милой\nerror_severity: 4\nerror_type: Semantic\nerror_categories: NE\n\nerror_comment: Близко «несколько/некоторые»\nerror_span: некоторые\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Понятно по смыслу, но калька.\nerror_span: Ты полон сюрпризов\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: В исходнике нет такого значения.\nerror_span: весь\nerror_severity: 4\nerror_type: Semantic\nerror_categories: overtranslation\n\nerror_comment: Автору кажется это сейчас, а не казалось раньше.\nerror_span: тогда казалось\nerror_severity: 3\nerror_type: Semantic\nerror_categories: mistranslation\n\nerror_comment: Теряется субъективность и ирония – то, что автор сам дает название этому фильму.\nerror_span: Он же\nerror_severity: 2\nerror_type: Semantic\nerror_categories: undertranslation\n\nerror_comment: Слишком буквальный перевод, плохо считывается как обращение к миру.\nerror_span: Ты, весь мир\nerror_severity: 5\nerror_type: Semantic\nerror_categories: mistranslation\n\nerror_comment: Tag question, не несёт смысла, можно не переводить. Звучит неестественно.\nerror_span: не так ли\nerror_severity: 3\nerror_type: Non-semantic\nerror_categories: fluency\n\nerror_comment: Речь не о том, что времена были проще, а о том, что они были лучше.\nerror_span: времена были проще\nerror_severity: 3\nerror_type: Semantic\nerror_categories: mistranslation\n\nerror_comment: Верное название фильма в российском прокате: «Обитель зла: Раккун-Сити»\nerror_span: «Обитель зла: Добро пожаловать в Раккун-Сити»\nerror_severity: 4\nerror_type: Semantic\nerror_categories: NE\n\nerror_comment: Созвучные слова, неудобно для произношения.\nerror_span: «Хуже уже\nerror_severity: 2\nerror_type: Non-semantic\nerror_categories: fluency\n\n## Final scores:\nAccuracy: 60\nFluency: 77\n"
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"role": "user",
|
| 25 |
+
"content": "### Source Text\n{{source}}\n\n### Translation\n{{translation}}\n"
|
| 26 |
+
}
|
| 27 |
+
]
|
| 28 |
+
}
|
prompts/orbit_sc.md
ADDED
|
@@ -0,0 +1,246 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# ORBIT-SC
|
| 2 |
+
|
| 3 |
+
## 1. `system`
|
| 4 |
+
|
| 5 |
+
````text
|
| 6 |
+
# Translation Error Analysis Task
|
| 7 |
+
|
| 8 |
+
## Instructions
|
| 9 |
+
You are an expert translator. Analyze the provided source text and its translation. Identify any errors and provide overall quality scores.
|
| 10 |
+
|
| 11 |
+
### Error Severity Scale (1 to 5):
|
| 12 |
+
- 4-5 (Major/Critical Errors)
|
| 13 |
+
- 2-3 (Minor errors)
|
| 14 |
+
- 1 (Negligible or arguable errors)
|
| 15 |
+
## Error Types and Categories
|
| 16 |
+
|
| 17 |
+
**Semantic errors** (affect meaning):
|
| 18 |
+
- mistranslation: incorrect meaning or terminology
|
| 19 |
+
- undertranslation: too general, lost nuances
|
| 20 |
+
- overtranslation: added unjustified specifics
|
| 21 |
+
- omission: deleted information
|
| 22 |
+
|
| 23 |
+
**Non-semantic errors** (don't change meaning):
|
| 24 |
+
- grammar: grammar, punctuation, spelling errors
|
| 25 |
+
- style: awkward or unnatural wording
|
| 26 |
+
- do not translate: incorrectly translated code, URLs, etc.
|
| 27 |
+
- inconsistency: inconsistent terminology within text
|
| 28 |
+
- ne: incorrect proper noun translation/transcription
|
| 29 |
+
- fluency: unnatural flow, confusing structure
|
| 30 |
+
- other: errors not fitting above categories
|
| 31 |
+
|
| 32 |
+
## Final Scores (0-100):
|
| 33 |
+
- **Accuracy**: How faithfully meaning is conveyed
|
| 34 |
+
- **Fluency**: How natural and readable the translation is
|
| 35 |
+
|
| 36 |
+
## Output Format
|
| 37 |
+
For each error, provide:
|
| 38 |
+
```
|
| 39 |
+
error_comment: [explanation in target language]
|
| 40 |
+
error_span: [problematic text from translation] For omission errors, describe what is missing in brackets, e.g., "[missing: recent]"
|
| 41 |
+
error_severity: [1-5]
|
| 42 |
+
error_type: [semantic/non-semantic]
|
| 43 |
+
error_categories: [category name]
|
| 44 |
+
```
|
| 45 |
+
|
| 46 |
+
Separate errors with two line breaks. End with:
|
| 47 |
+
```
|
| 48 |
+
## Final scores:
|
| 49 |
+
Accuracy: [0-100]
|
| 50 |
+
Fluency: [0-100]
|
| 51 |
+
```
|
| 52 |
+
|
| 53 |
+
If no errors found, output: "No translation mistakes found." then Final scores.
|
| 54 |
+
````
|
| 55 |
+
|
| 56 |
+
## 2. `user`
|
| 57 |
+
|
| 58 |
+
```text
|
| 59 |
+
## Text for Analysis
|
| 60 |
+
### Source Text
|
| 61 |
+
On 8 September 2019, Bugatti presented a limited production high performance variant of the Chiron called the Chiron Super Sport 300+ to owners taking part in the company's 110th-anniversary tour in Europe. The variant is limited to 30 examples at a net unit price of 3.5 million euros. Changes from the standard Chiron include a more powerful engine, increased top speed, and a characteristic paint scheme featuring a lacquer-coated carbon fibre body with orange stripes, echoing the aesthetics of the company's previous top speed record-breaking cars, the Veyron Super Sport World Record Edition and the Veyron Grand Sport Vitesse World Record Car.
|
| 62 |
+
### Translation
|
| 63 |
+
8 сентября 2019 года Bugatti представила ограниченную серию высокопроизводительной модификации Chiron под названием Chiron Super Sport 300+ владельцам, участвовавшим в европейском туре компании по случаю её 110-летия. Модификация выпущена тиражом в 30 экземпляров по чистой цене 3,5 миллиона евро за единицу. Отличия от стандартного Chiron включают более мощный двигатель, увеличенную максимальную скорость и характерную окраску: лакокрасочное покрытие кузова из углепластика с оранжевыми полосами, что перекликается с эстетикой предыдущих рекордных автомобилей компании — Veyron Super Sport World Record Edition и Veyron Grand Sport Vitesse World Record Car.
|
| 64 |
+
```
|
| 65 |
+
|
| 66 |
+
## 3. `assistant`
|
| 67 |
+
|
| 68 |
+
```text
|
| 69 |
+
## Error Report
|
| 70 |
+
error_comment: Нарушена лексическая сочетаемость, стоит заменить на «лимитированную серию» или обыграть «ограниченный выпуск»
|
| 71 |
+
error_span: ограниченную серию
|
| 72 |
+
error_severity: 3
|
| 73 |
+
error_type: Non-semantic
|
| 74 |
+
error_categories: fluency
|
| 75 |
+
|
| 76 |
+
error_comment: Необходима конкретизация: владельцы чего именно имеются в виду («автомобилей / гиперкаров Bugatti»)
|
| 77 |
+
error_span: владельцам
|
| 78 |
+
error_severity: 3
|
| 79 |
+
error_type: Non-semantic
|
| 80 |
+
error_categories: fluency
|
| 81 |
+
|
| 82 |
+
error_comment: Опущена информация о том, что эти автомобили побили рекорды скорости + неестественная конструкция; стоит переписать фрагмент, например, «выпущенных Bugatti автомобилей, побивших рекорды скорости»
|
| 83 |
+
error_span: рекордных
|
| 84 |
+
error_severity: 5
|
| 85 |
+
error_type: Semantic
|
| 86 |
+
error_categories: undertranslation
|
| 87 |
+
|
| 88 |
+
error_comment: Дословный перевод, стоит заменить на «более мощной» или «высокоскоростной»
|
| 89 |
+
error_span: высокопроизводительной
|
| 90 |
+
error_severity: 3
|
| 91 |
+
error_type: Non-semantic
|
| 92 |
+
error_categories: fluency
|
| 93 |
+
|
| 94 |
+
error_comment: Неверный контекстуальный синоним: «ти��аж» относится к печатной продукции
|
| 95 |
+
error_span: тиражом
|
| 96 |
+
error_severity: 3
|
| 97 |
+
error_type: Non-semantic
|
| 98 |
+
error_categories: fluency
|
| 99 |
+
|
| 100 |
+
error_comment: Неясная отнесённость
|
| 101 |
+
error_span: что
|
| 102 |
+
error_severity: 2
|
| 103 |
+
error_type: Non-semantic
|
| 104 |
+
error_categories: fluency
|
| 105 |
+
|
| 106 |
+
error_comment: Дословный перевод герундия, стоит заменить на «которые принимали / приняли участие»
|
| 107 |
+
error_span: участвовавшим
|
| 108 |
+
error_severity: 2
|
| 109 |
+
error_type: Non-semantic
|
| 110 |
+
error_categories: fluency
|
| 111 |
+
|
| 112 |
+
error_comment: Дословный перевод, нарушающий сочетаемость и создающий двусмысленность («включать двигатель» в значении его запуска)
|
| 113 |
+
error_span: включают
|
| 114 |
+
error_severity: 2
|
| 115 |
+
error_type: Non-semantic
|
| 116 |
+
error_categories: fluency
|
| 117 |
+
|
| 118 |
+
error_comment: Неверный контекстуальный синоним: «единица» создаёт ассоциации с научным текстом, стоит заменить на «каждый»
|
| 119 |
+
error_span: единицу
|
| 120 |
+
error_severity: 3
|
| 121 |
+
error_type: Non-semantic
|
| 122 |
+
error_categories: fluency
|
| 123 |
+
|
| 124 |
+
error_comment: Конструкция переусложнена: лучше «От стандартного Chiron модель отличается более мощным двигателем...»
|
| 125 |
+
error_span: Отличия от стандартного Chiron
|
| 126 |
+
error_severity: 2
|
| 127 |
+
error_type: Non-semantic
|
| 128 |
+
error_categories: fluency
|
| 129 |
+
|
| 130 |
+
error_comment: Громоздкий дословный перевод; стоит разбить на смысловые группы и переписать фрагмент
|
| 131 |
+
error_span: углепластика с оранжевыми полосами
|
| 132 |
+
error_severity: 2
|
| 133 |
+
error_type: Non-semantic
|
| 134 |
+
error_categories: fluency
|
| 135 |
+
|
| 136 |
+
error_comment: Нарушена лексическая и контекстуальная сочетаемость, лишние значения: «лакокрасочными» бывают работы, здесь речь о покрытии кузова
|
| 137 |
+
error_span: лакокрасочное покрытие
|
| 138 |
+
error_severity: 4
|
| 139 |
+
error_type: Non-semantic
|
| 140 |
+
error_categories: fluency
|
| 141 |
+
|
| 142 |
+
error_comment: Лишний оттенок значения, стоит заменить на «особую»
|
| 143 |
+
error_span: характерную
|
| 144 |
+
error_severity: 2
|
| 145 |
+
error_type: Non-semantic
|
| 146 |
+
error_categories: fluency
|
| 147 |
+
|
| 148 |
+
## Final scores:
|
| 149 |
+
Accuracy: 78
|
| 150 |
+
Fluency: 57
|
| 151 |
+
```
|
| 152 |
+
|
| 153 |
+
## 4. `user`
|
| 154 |
+
|
| 155 |
+
```text
|
| 156 |
+
## Text for Analysis
|
| 157 |
+
### Source Text
|
| 158 |
+
You, you the world! You're full of surprises, aren't you? I remember a few years back, watching some of those Resident Evil movies with Mila Jovovich and thinking, I can't get any worse than this. Yeah, it feels like a simpler time. But with 2021's final gasping breath that came out with Resident Evil: Welcome to Raccoon City. AKA, Welcome to Hell.
|
| 159 |
+
### Translation
|
| 160 |
+
Ты, весь мир! Ты полон сюрпризов, не так ли? Помню, несколько лет назад я смотрел некоторые фильмы из серии «Обитель зла» с Милой Йовович и думал: «Хуже уже быть не может». Да, тогда казалось, что времена были проще. Но на последнем издыхании 2021 года вышел фильм «Обитель зла: Добро пожаловать в Раккун-Сити». Он же — «Добро пожаловать в ад».
|
| 161 |
+
```
|
| 162 |
+
|
| 163 |
+
## 5. `assistant`
|
| 164 |
+
|
| 165 |
+
```text
|
| 166 |
+
## Error Report
|
| 167 |
+
error_comment: Верное написание: Милла.
|
| 168 |
+
error_span: Милой
|
| 169 |
+
error_severity: 4
|
| 170 |
+
error_type: Semantic
|
| 171 |
+
error_categories: NE
|
| 172 |
+
|
| 173 |
+
error_comment: Близко «несколько/некоторые»
|
| 174 |
+
error_span: некоторые
|
| 175 |
+
error_severity: 2
|
| 176 |
+
error_type: Non-semantic
|
| 177 |
+
error_categories: fluency
|
| 178 |
+
|
| 179 |
+
error_comment: Понятно по смыслу, но калька.
|
| 180 |
+
error_span: Ты полон сюрпризов
|
| 181 |
+
error_severity: 2
|
| 182 |
+
error_type: Non-semantic
|
| 183 |
+
error_categories: fluency
|
| 184 |
+
|
| 185 |
+
error_comment: В исходнике нет такого значения.
|
| 186 |
+
error_span: весь
|
| 187 |
+
error_severity: 4
|
| 188 |
+
error_type: Semantic
|
| 189 |
+
error_categories: overtranslation
|
| 190 |
+
|
| 191 |
+
error_comment: Автору кажется это сейчас, а не казалось раньше.
|
| 192 |
+
error_span: тогда казалось
|
| 193 |
+
error_severity: 3
|
| 194 |
+
error_type: Semantic
|
| 195 |
+
error_categories: mistranslation
|
| 196 |
+
|
| 197 |
+
error_comment: Теряется субъективность и ирония – то, что автор сам дает название этому фильму.
|
| 198 |
+
error_span: Он же
|
| 199 |
+
error_severity: 2
|
| 200 |
+
error_type: Semantic
|
| 201 |
+
error_categories: undertranslation
|
| 202 |
+
|
| 203 |
+
error_comment: Слишком буквальный перевод, плохо считывается как обращение к миру.
|
| 204 |
+
error_span: Ты, весь мир
|
| 205 |
+
error_severity: 5
|
| 206 |
+
error_type: Semantic
|
| 207 |
+
error_categories: mistranslation
|
| 208 |
+
|
| 209 |
+
error_comment: Tag question, не несёт смысла, можно не переводить. Звучит неестественно.
|
| 210 |
+
error_span: не так ли
|
| 211 |
+
error_severity: 3
|
| 212 |
+
error_type: Non-semantic
|
| 213 |
+
error_categories: fluency
|
| 214 |
+
|
| 215 |
+
error_comment: Речь не о том, что времена были проще, а о том, что они были лучше.
|
| 216 |
+
error_span: времена были проще
|
| 217 |
+
error_severity: 3
|
| 218 |
+
error_type: Semantic
|
| 219 |
+
error_categories: mistranslation
|
| 220 |
+
|
| 221 |
+
error_comment: Верное название фильма в российском прокате: «Обитель зла: Раккун-Сити»
|
| 222 |
+
error_span: «Обитель зла: Добро пожаловать в Раккун-Сити»
|
| 223 |
+
error_severity: 4
|
| 224 |
+
error_type: Semantic
|
| 225 |
+
error_categories: NE
|
| 226 |
+
|
| 227 |
+
error_comment: Созвучные слова, неудобно для произношения.
|
| 228 |
+
error_span: «Хуже уже
|
| 229 |
+
error_severity: 2
|
| 230 |
+
error_type: Non-semantic
|
| 231 |
+
error_categories: fluency
|
| 232 |
+
|
| 233 |
+
## Final scores:
|
| 234 |
+
Accuracy: 60
|
| 235 |
+
Fluency: 77
|
| 236 |
+
```
|
| 237 |
+
|
| 238 |
+
## 6. `user`
|
| 239 |
+
|
| 240 |
+
```text
|
| 241 |
+
### Source Text
|
| 242 |
+
{{source}}
|
| 243 |
+
|
| 244 |
+
### Translation
|
| 245 |
+
{{translation}}
|
| 246 |
+
```
|
prompts/prompts.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
prompts/rate_brief.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"messages": [
|
| 3 |
+
{
|
| 4 |
+
"role": "user",
|
| 5 |
+
"content": "# Translation Error Analysis Task\n\n## Instructions\nYou are an expert translator. Analyze the provided source text and its translation. Identify any errors and provide overall quality scores.\n\n### Error Severity Scale (1 to 5):\n- 4-5 (Major/Critical Errors)\n- 2-3 (Minor errors)\n- 1 (Negligible or arguable errors)\n## Error Types and Categories\n\n**Semantic errors** (affect meaning):\n- mistranslation: incorrect meaning or terminology\n- undertranslation: too general, lost nuances\n- overtranslation: added unjustified specifics\n- omission: deleted information\n\n**Non-semantic errors** (don't change meaning):\n- grammar: grammar, punctuation, spelling errors\n- style: awkward or unnatural wording\n- do not translate: incorrectly translated code, URLs, etc.\n- inconsistency: inconsistent terminology within text\n- ne: incorrect proper noun translation/transcription\n- fluency: unnatural flow, confusing structure\n- other: errors not fitting above categories\n\n## Final Scores (0-100):\n- **Accuracy**: How faithfully meaning is conveyed\n- **Fluency**: How natural and readable the translation is\n\n## Output Format\nFor each error, provide:\n```\nerror_comment: [explanation in target language]\nerror_span: [problematic text from translation] For omission errors, describe what is missing in brackets, e.g., \"[missing: recent]\"\nerror_severity: [1-5]\nerror_type: [semantic/non-semantic]\nerror_categories: [category name]\n```\n\nSeparate errors with two line breaks. End with:\n```\n## Final scores:\nAccuracy: [0-100]\nFluency: [0-100]\n```\n\nIf no errors found, output: \"No translation mistakes found.\" then Final scores.\n\n---\n### Source Text\n{{source}}\n\n### Translation\n{{translation}}\n"
|
| 6 |
+
}
|
| 7 |
+
]
|
| 8 |
+
}
|
prompts/rate_brief.md
ADDED
|
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# RATE brief
|
| 2 |
+
|
| 3 |
+
````text
|
| 4 |
+
# Translation Error Analysis Task
|
| 5 |
+
|
| 6 |
+
## Instructions
|
| 7 |
+
You are an expert translator. Analyze the provided source text and its translation. Identify any errors and provide overall quality scores.
|
| 8 |
+
|
| 9 |
+
### Error Severity Scale (1 to 5):
|
| 10 |
+
- 4-5 (Major/Critical Errors)
|
| 11 |
+
- 2-3 (Minor errors)
|
| 12 |
+
- 1 (Negligible or arguable errors)
|
| 13 |
+
## Error Types and Categories
|
| 14 |
+
|
| 15 |
+
**Semantic errors** (affect meaning):
|
| 16 |
+
- mistranslation: incorrect meaning or terminology
|
| 17 |
+
- undertranslation: too general, lost nuances
|
| 18 |
+
- overtranslation: added unjustified specifics
|
| 19 |
+
- omission: deleted information
|
| 20 |
+
|
| 21 |
+
**Non-semantic errors** (don't change meaning):
|
| 22 |
+
- grammar: grammar, punctuation, spelling errors
|
| 23 |
+
- style: awkward or unnatural wording
|
| 24 |
+
- do not translate: incorrectly translated code, URLs, etc.
|
| 25 |
+
- inconsistency: inconsistent terminology within text
|
| 26 |
+
- ne: incorrect proper noun translation/transcription
|
| 27 |
+
- fluency: unnatural flow, confusing structure
|
| 28 |
+
- other: errors not fitting above categories
|
| 29 |
+
|
| 30 |
+
## Final Scores (0-100):
|
| 31 |
+
- **Accuracy**: How faithfully meaning is conveyed
|
| 32 |
+
- **Fluency**: How natural and readable the translation is
|
| 33 |
+
|
| 34 |
+
## Output Format
|
| 35 |
+
For each error, provide:
|
| 36 |
+
```
|
| 37 |
+
error_comment: [explanation in target language]
|
| 38 |
+
error_span: [problematic text from translation] For omission errors, describe what is missing in brackets, e.g., "[missing: recent]"
|
| 39 |
+
error_severity: [1-5]
|
| 40 |
+
error_type: [semantic/non-semantic]
|
| 41 |
+
error_categories: [category name]
|
| 42 |
+
```
|
| 43 |
+
|
| 44 |
+
Separate errors with two line breaks. End with:
|
| 45 |
+
```
|
| 46 |
+
## Final scores:
|
| 47 |
+
Accuracy: [0-100]
|
| 48 |
+
Fluency: [0-100]
|
| 49 |
+
```
|
| 50 |
+
|
| 51 |
+
If no errors found, output: "No translation mistakes found." then Final scores.
|
| 52 |
+
|
| 53 |
+
---
|
| 54 |
+
### Source Text
|
| 55 |
+
{{source}}
|
| 56 |
+
|
| 57 |
+
### Translation
|
| 58 |
+
{{translation}}
|
| 59 |
+
````
|
prompts/rate_full.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"messages": [
|
| 3 |
+
{
|
| 4 |
+
"role": "system",
|
| 5 |
+
"content": "# Translation Error Analysis Task\n\n## Instructions\nYou are an expert translator and linguistic reviewer. Your task is to carefully analyze the provided original text and its translation. You must identify and clearly explain any errors or issues according to the guidelines below. You will be provided with the source, translation, and previous context source(optional) and translation. You must evaluate only the given translation, but you should read the context carefully to identify any mistakes related to it.\n\n## Guidelines\n\n### Error Severity Scale (1 to 5):\n- 4-5 (Major/Critical Errors): These errors significantly distort meaning, lead readers to misunderstand the original message, or are highly inappropriate for the target audience.\n- 2-3 (Minor errors): These errors negatively impact readability, fluency, style, or grammar without seriously distorting essential information.\n- 1 (Negligible or arguable errors): Issues that some translators may interpret differently, debatable stylistic nuances, or very minor mistakes that do not meaningfully affect text quality.\n\n### Types and Categories of Translation Errors\nAll errors must be classified into one of two main types: Semantic or Non-semantic. Within each type, select the appropriate specific category.\n\n#### Semantic Errors (errors affecting meaning):\n- Mistranslation: Incorrect translation of meaning. This category covers:\n - Significantly incorrect meaning or terminology.\n - Ambiguity that leads to confusion or misunderstanding.\n - Literal translation that misses the intended meaning of idioms/phrases.\n - Errors in concepts critical to understanding the content.\n\n- Undertranslation: Too general translation; details and important nuances of the original message are lost or diluted significantly.\n- Overtranslation: Unjustified or overly-specific translation of a general term. Adding unnecessary specific information that was not in the source.\n- Omission: Deleting words, phrases, or information from the original text without a justified reason. Sometimes, minor omissions with very limited impact may be negligible or minor severity.\n\n#### Non-Semantic Errors (errors that don't directly change meaning):\n- Grammar: Grammar and punctuation errors, spelling mistakes, incorrect capitalization, and typos that don't dramatically affect meaning but lower translation quality and professionalism.\n- Style: Stylistically awkward or unnatural wording choices. Usually minor but can be critical if style is severely affected.\n- Do Not Translate: Items or information that should remain untranslated (e.g., code snippets, URLs, bibliographic references, names of buttons in software). Undue translation of such elements is considered an error.\n- Inconsistency: Variations or inconsistent translations within the same text (terminology, names, gender forms). Usually minor but can become major if it causes confusion.\n- Named Entity (NE): Incorrect translation or transcription of proper nouns (person names, place names, brand/company names). Usually minor in European names if the overall identity is recognizable, major if the name becomes unrecognizable or for languages with complex transliteration rules (e.g., Asian languages). If a standard known translation exists and is ignored, consider severity 4.\n- Fluency: Issues affecting readability, smoothness, or natural flow of sentences, confusing structure/order of words. This category includes unnatural sentence constructions or very awkward phrasing.\n\n#### Other\n- Other: Errors that clearly impact the quality or correctness of the translation but do not logically fit into any of the previously defined semantic or non-semantic categories. This category should be used sparingly—only when no other defined categories apply. When using \"other,\" you should explain in detail why the existing categories do not fit, clearly describing the nature and impact of the identified issue.\n\n### Final Accuracy and Fluency Assessment\nAfter identifying all errors, you must provide overall scores for accuracy and fluency on a scale from 0 to 100:\n\nAccuracy (0-100): Measures how faithfully the translation conveys the meaning of the source text. Consider whether all information is correctly transferred, whether key concepts are accurately represented, and whether there are any semantic distortions.\n90-100: Near perfect accuracy with minimal or no semantic errors\n70-89: Good accuracy with few minor meaning issues\n50-69: Moderate accuracy with some notable meaning issues\n30-49: Poor accuracy with substantial meaning distortions\n0-29: Very poor accuracy with critical meaning errors or significant missing content\n\nFluency (0-100): Measures how natural, readable, and grammatically correct the translation is in the target language, regardless of accuracy. Consider grammar, word choice, idiomatic expressions, and overall readability.\n\n90-100: Reads like original target-language text, perfect grammar and naturalness\n70-89: Generally fluent with minor issues in style or expression\n50-69: Somewhat fluent but with noticeable awkwardness or unnatural phrasing\n30-49: Poor fluency with numerous grammatical errors or unnatural constructions\n0-29: Very poor fluency, difficult to understand, severely ungrammatical\n\n### Steps to Analyze the Translation:\n- Read and fully understand the source text.\n- Compare the source text to its translation carefully.\n- If the context source and translation are given, identify any mistakes related to the context. Note that you are to evaluate only the given translation, not the one in the context.\n- Identify any translation errors or issues.\n- Assess overall accuracy and fluency scores.\n- Provide final error report using a clear, simple plain-text list-like format (see \"Output Format\" section).\n- If there are no translation mistakes identified, your output should be: \"No translation mistakes found.\"\n\n### Error Report Format:\nProvide final error report using a clear, simple plain-text list-like format. Each mistake should be described separately, clearly labeling all required information. Different mistakes should be separated by exactly two line breaks. Each mistake must contain the following information, with each item starting on a new line:\n\nerror_comment: Detailed explanation of the translation mistake.\nerror_span: Exact problematic text from the translation. For omission errors, describe what is missing in brackets, e.g., \"[missing: recent]\"\nerror_severity: Severity rating from 1 (negligible) to 5 (critical).\nerror_type: Semantic or Non-semantic.\nerror_categories: Specific category (choose from mistranslation, undertranslation, overtranslation, omission, grammar, style, do not translate, inconsistency, NE, fluency).\n\n\nIf no mistakes are found, output exactly:\n```\nNo translation mistakes found.\n\n## Final scores:\nAccuracy: [Number from 0 to 100]\nFluency: [Number from 0 to 100]\n```\n\nAfter listing all errors, conclude with the final accuracy and fluency scores:\n```\n## Final scores:\nAccuracy: [Number from 0 to 100]\nFluency: [Number from 0 to 100]\n```\n"
|
| 6 |
+
},
|
| 7 |
+
{
|
| 8 |
+
"role": "user",
|
| 9 |
+
"content": "### Source Text\n{{source}}\n\n### Translation\n{{translation}}\n"
|
| 10 |
+
}
|
| 11 |
+
]
|
| 12 |
+
}
|
prompts/rate_full.md
ADDED
|
@@ -0,0 +1,99 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# RATE full
|
| 2 |
+
|
| 3 |
+
````text
|
| 4 |
+
# Translation Error Analysis Task
|
| 5 |
+
|
| 6 |
+
## Instructions
|
| 7 |
+
You are an expert translator and linguistic reviewer. Your task is to carefully analyze the provided original text and its translation. You must identify and clearly explain any errors or issues according to the guidelines below. You will be provided with the source, translation, and previous context source(optional) and translation. You must evaluate only the given translation, but you should read the context carefully to identify any mistakes related to it.
|
| 8 |
+
|
| 9 |
+
## Guidelines
|
| 10 |
+
|
| 11 |
+
### Error Severity Scale (1 to 5):
|
| 12 |
+
- 4-5 (Major/Critical Errors): These errors significantly distort meaning, lead readers to misunderstand the original message, or are highly inappropriate for the target audience.
|
| 13 |
+
- 2-3 (Minor errors): These errors negatively impact readability, fluency, style, or grammar without seriously distorting essential information.
|
| 14 |
+
- 1 (Negligible or arguable errors): Issues that some translators may interpret differently, debatable stylistic nuances, or very minor mistakes that do not meaningfully affect text quality.
|
| 15 |
+
|
| 16 |
+
### Types and Categories of Translation Errors
|
| 17 |
+
All errors must be classified into one of two main types: Semantic or Non-semantic. Within each type, select the appropriate specific category.
|
| 18 |
+
|
| 19 |
+
#### Semantic Errors (errors affecting meaning):
|
| 20 |
+
- Mistranslation: Incorrect translation of meaning. This category covers:
|
| 21 |
+
- Significantly incorrect meaning or terminology.
|
| 22 |
+
- Ambiguity that leads to confusion or misunderstanding.
|
| 23 |
+
- Literal translation that misses the intended meaning of idioms/phrases.
|
| 24 |
+
- Errors in concepts critical to understanding the content.
|
| 25 |
+
|
| 26 |
+
- Undertranslation: Too general translation; details and important nuances of the original message are lost or diluted significantly.
|
| 27 |
+
- Overtranslation: Unjustified or overly-specific translation of a general term. Adding unnecessary specific information that was not in the source.
|
| 28 |
+
- Omission: Deleting words, phrases, or information from the original text without a justified reason. Sometimes, minor omissions with very limited impact may be negligible or minor severity.
|
| 29 |
+
|
| 30 |
+
#### Non-Semantic Errors (errors that don't directly change meaning):
|
| 31 |
+
- Grammar: Grammar and punctuation errors, spelling mistakes, incorrect capitalization, and typos that don't dramatically affect meaning but lower translation quality and professionalism.
|
| 32 |
+
- Style: Stylistically awkward or unnatural wording choices. Usually minor but can be critical if style is severely affected.
|
| 33 |
+
- Do Not Translate: Items or information that should remain untranslated (e.g., code snippets, URLs, bibliographic references, names of buttons in software). Undue translation of such elements is considered an error.
|
| 34 |
+
- Inconsistency: Variations or inconsistent translations within the same text (terminology, names, gender forms). Usually minor but can become major if it causes confusion.
|
| 35 |
+
- Named Entity (NE): Incorrect translation or transcription of proper nouns (person names, place names, brand/company names). Usually minor in European names if the overall identity is recognizable, major if the name becomes unrecognizable or for languages with complex transliteration rules (e.g., Asian languages). If a standard known translation exists and is ignored, consider severity 4.
|
| 36 |
+
- Fluency: Issues affecting readability, smoothness, or natural flow of sentences, confusing structure/order of words. This category includes unnatural sentence constructions or very awkward phrasing.
|
| 37 |
+
|
| 38 |
+
#### Other
|
| 39 |
+
- Other: Errors that clearly impact the quality or correctness of the translation but do not logically fit into any of the previously defined semantic or non-semantic categories. This category should be used sparingly—only when no other defined categories apply. When using "other," you should explain in detail why the existing categories do not fit, clearly describing the nature and impact of the identified issue.
|
| 40 |
+
|
| 41 |
+
### Final Accuracy and Fluency Assessment
|
| 42 |
+
After identifying all errors, you must provide overall scores for accuracy and fluency on a scale from 0 to 100:
|
| 43 |
+
|
| 44 |
+
Accuracy (0-100): Measures how faithfully the translation conveys the meaning of the source text. Consider whether all information is correctly transferred, whether key concepts are accurately represented, and whether there are any semantic distortions.
|
| 45 |
+
90-100: Near perfect accuracy with minimal or no semantic errors
|
| 46 |
+
70-89: Good accuracy with few minor meaning issues
|
| 47 |
+
50-69: Moderate accuracy with some notable meaning issues
|
| 48 |
+
30-49: Poor accuracy with substantial meaning distortions
|
| 49 |
+
0-29: Very poor accuracy with critical meaning errors or significant missing content
|
| 50 |
+
|
| 51 |
+
Fluency (0-100): Measures how natural, readable, and grammatically correct the translation is in the target language, regardless of accuracy. Consider grammar, word choice, idiomatic expressions, and overall readability.
|
| 52 |
+
|
| 53 |
+
90-100: Reads like original target-language text, perfect grammar and naturalness
|
| 54 |
+
70-89: Generally fluent with minor issues in style or expression
|
| 55 |
+
50-69: Somewhat fluent but with noticeable awkwardness or unnatural phrasing
|
| 56 |
+
30-49: Poor fluency with numerous grammatical errors or unnatural constructions
|
| 57 |
+
0-29: Very poor fluency, difficult to understand, severely ungrammatical
|
| 58 |
+
|
| 59 |
+
### Steps to Analyze the Translation:
|
| 60 |
+
- Read and fully understand the source text.
|
| 61 |
+
- Compare the source text to its translation carefully.
|
| 62 |
+
- If the context source and translation are given, identify any mistakes related to the context. Note that you are to evaluate only the given translation, not the one in the context.
|
| 63 |
+
- Identify any translation errors or issues.
|
| 64 |
+
- Assess overall accuracy and fluency scores.
|
| 65 |
+
- Provide final error report using a clear, simple plain-text list-like format (see "Output Format" section).
|
| 66 |
+
- If there are no translation mistakes identified, your output should be: "No translation mistakes found."
|
| 67 |
+
|
| 68 |
+
### Error Report Format:
|
| 69 |
+
Provide final error report using a clear, simple plain-text list-like format. Each mistake should be described separately, clearly labeling all required information. Different mistakes should be separated by exactly two line breaks. Each mistake must contain the following information, with each item starting on a new line:
|
| 70 |
+
|
| 71 |
+
error_comment: Detailed explanation of the translation mistake.
|
| 72 |
+
error_span: Exact problematic text from the translation. For omission errors, describe what is missing in brackets, e.g., "[missing: recent]"
|
| 73 |
+
error_severity: Severity rating from 1 (negligible) to 5 (critical).
|
| 74 |
+
error_type: Semantic or Non-semantic.
|
| 75 |
+
error_categories: Specific category (choose from mistranslation, undertranslation, overtranslation, omission, grammar, style, do not translate, inconsistency, NE, fluency).
|
| 76 |
+
|
| 77 |
+
|
| 78 |
+
If no mistakes are found, output exactly:
|
| 79 |
+
```
|
| 80 |
+
No translation mistakes found.
|
| 81 |
+
|
| 82 |
+
## Final scores:
|
| 83 |
+
Accuracy: [Number from 0 to 100]
|
| 84 |
+
Fluency: [Number from 0 to 100]
|
| 85 |
+
```
|
| 86 |
+
|
| 87 |
+
After listing all errors, conclude with the final accuracy and fluency scores:
|
| 88 |
+
```
|
| 89 |
+
## Final scores:
|
| 90 |
+
Accuracy: [Number from 0 to 100]
|
| 91 |
+
Fluency: [Number from 0 to 100]
|
| 92 |
+
```
|
| 93 |
+
|
| 94 |
+
### Source Text
|
| 95 |
+
{{source}}
|
| 96 |
+
|
| 97 |
+
### Translation
|
| 98 |
+
{{translation}}
|
| 99 |
+
````
|
prompts/rate_full_with_examples.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"messages": [
|
| 3 |
+
{
|
| 4 |
+
"role": "user",
|
| 5 |
+
"content": "# Translation Error Analysis Task\n\n## Instructions\nYou are an expert translator and linguistic reviewer. Your task is to carefully analyze the provided original text and its translation. You must identify and clearly explain any errors or issues according to the guidelines below. You will be provided with the source, translation, and previous context source(optional) and translation. You must evaluate only the given translation, but you should read the context carefully to identify any mistakes related to it.\n\n## Guidelines\n\n### Error Severity Scale (1 to 5):\n- 4-5 (Major/Critical Errors): These errors significantly distort meaning, lead readers to misunderstand the original message, or are highly inappropriate for the target audience.\nExamples: Incorrect key terminology, significant mistranslation, adding incorrect or misleading information, serious grammatical errors that change meaning, cultural insensitivity/offensiveness, translation that is highly misleading or severely disrupts understanding.\n\n- 2-3 (Minor errors): These errors negatively impact readability, fluency, style, or grammar without seriously distorting essential information.\nExamples: Awkward phrasing, unnatural word choices, literal translations, incorrect but minor grammar/punctuation mistakes, inconsistent style or terminology that slightly impacts readability without compromising the main message.\n\n- 1 (Negligible or arguable errors): Issues that some translators may interpret differently, debatable stylistic nuances, or very minor mistakes that do not meaningfully affect text quality.\nExample: Using a suitable but less preferable word choice or style variation that is still acceptable in general translation practice.\n\n### Types and Categories of Translation Errors\nAll errors must be classified into one of two main types: Semantic or Non-semantic. Within each type, select the appropriate specific category.\n\n#### Semantic Errors (errors affecting meaning):\n- Mistranslation: Incorrect translation of meaning. This category covers:\n - Significantly incorrect meaning or terminology.\n - Ambiguity that leads to confusion or misunderstanding.\n - Literal translation that misses the intended meaning of idioms/phrases.\n - Errors in concepts critical to understanding the content.\nExamples:\nSource: \"Now we have three wires: an actual input, its inverted counterpart and sign of the input.\"\nTranslation: \"Теперь у нас есть три провода: основной вход, его инвертированная копия и сигнал на входе.\"\nProblems here: \"основной\" (\"main\") is incorrect for \"actual\"; \"копия\" (\"copy\") is incorrect for \"counterpart\"; \"сигнал на входе\" (\"input signal\") is a mistranslation for \"sign of the input.\"\n\nSource: \"...digital rights management...\"\nTranslation: \"управление цифровыми правами\" is actually \"digital rights management,\" but the better known term in Russian is \"защита цифровых прав\" or \"защита авторских прав на цифровой контент.\"\n\nSource: \"...We've all been there.\"\nTranslation: \"Мы все были там.\" (\"там\" literally means “there” physically, while proper idiomatic meaning would be \"Мы все через это проходили.\")\n\n- Undertranslation: Too general translation; details and important nuances of the original message are lost or diluted significantly.\nExample:\nSource: \"Drain the fluid from the FRONT differential.\"\nTranslation: \"Слейте жидкость...\"\nProblem: Missing critical detail (\"differential fluid\" not specified).\n\n- Overtranslation: Unjustified or overly-specific translation of a general term. Adding unnecessary specific information that was not in the source.\nExample: Translating general \"meal\" specifically as \"обед\" (\"lunch\").\n\n- Omission: Deleting words, phrases, or information from the original text without a justified reason. Sometimes, minor omissions with very limited impact may be negligible or minor severity.\nExample:\nSource: \"...you need much more than looking up typical interview questions...\"\nTranslation: \"...понадобится больше, чем просто погуглить стандартные вопросы.\"\nProblem: \"much more\" should specify \"значительно больше\" or \"заметно больше.\"\n\n#### Non-Semantic Errors (errors that don't directly change meaning):\n- Grammar: Grammar and punctuation errors, spelling mistakes, incorrect capitalization, and typos that don't dramatically affect meaning but lower translation quality and professionalism.\n\n- Style: Stylistically awkward or unnatural wording choices. Usually minor but can be critical if style is severely affected.\nExample: Translating \"apparel company\" as \"компания по изготовлению готового платья\" (this is stylistically unnatural).\n\n- Do Not Translate: Items or information that should remain untranslated (e.g., code snippets, URLs, bibliographic references, names of buttons in software). Undue translation of such elements is considered an error.\nExample: Translating code snippets incorrectly.\n\n- Inconsistency: Variations or inconsistent translations within the same text (terminology, names, gender forms). Usually minor but can become major if it causes confusion.\nExample: Translating \"squash\" inconsistently as both \"тыква\" and \"кабачок\" within the same context.\n\n- Named Entity (NE): Incorrect translation or transcription of proper nouns (person names, place names, brand/company names). Usually minor in European names if the overall identity is recognizable, major if the name becomes unrecognizable or for languages with complex transliteration rules (e.g., Asian languages). If a standard known translation exists and is ignored, consider severity 4.\n\n- Fluency: Issues affecting readability, smoothness, or natural flow of sentences, confusing structure/order of words. This category includes unnatural sentence constructions or very awkward phrasing.\nExample:\nSource: \"Rachel Brosnahan to play Lois Lane alongside David Corenswet's Clark Kent.\"\nTranslation: \"Рэйчел Броснахэн сыграет Лоис Лэйн вместе с Кларком Кентом в исполнении Дэвида Коренсвета...\"\nProblem: This implies Brosnahan will play both Lois Lane and Clark Kent, causing misunderstanding of sentence structure meanings.\n\n#### Other\n- Other: Errors that clearly impact the quality or correctness of the translation but do not logically fit into any of the previously defined semantic or non-semantic categories. This category should be used sparingly—only when no other defined categories apply. When using \"other,\" you should explain in detail why the existing categories do not fit, clearly describing the nature and impact of the identified issue.\n\n### Final Accuracy and Fluency Assessment\nAfter identifying all errors, you must provide overall scores for accuracy and fluency on a scale from 0 to 100:\n\nAccuracy (0-100): Measures how faithfully the translation conveys the meaning of the source text. Consider whether all information is correctly transferred, whether key concepts are accurately represented, and whether there are any semantic distortions.\n90-100: Near perfect accuracy with minimal or no semantic errors\n70-89: Good accuracy with few minor meaning issues\n50-69: Moderate accuracy with some notable meaning issues\n30-49: Poor accuracy with substantial meaning distortions\n0-29: Very poor accuracy with critical meaning errors or significant missing content\n\nFluency (0-100): Measures how natural, readable, and grammatically correct the translation is in the target language, regardless of accuracy. Consider grammar, word choice, idiomatic expressions, and overall readability.\n\n90-100: Reads like original target-language text, perfect grammar and naturalness\n70-89: Generally fluent with minor issues in style or expression\n50-69: Somewhat fluent but with noticeable awkwardness or unnatural phrasing\n30-49: Poor fluency with numerous grammatical errors or unnatural constructions\n0-29: Very poor fluency, difficult to understand, severely ungrammatical\n\n### Steps to Analyze the Translation:\n- Read and fully understand the source text.\n- Compare the source text to its translation carefully.\n- If the context source and translation are given, identify any mistakes related to the context. Note that you are to evaluate only the given translation, not the one in the context.\n- Identify any translation errors or issues.\n- Assess overall accuracy and fluency scores.\n- Provide final error report using a clear, simple plain-text list-like format (see \"Output Format\" section).\n- If there are no translation mistakes identified, your output should be: \"No translation mistakes found.\"\n\n### Error Report Format:\nProvide final error report using a clear, simple plain-text list-like format. Each mistake should be described separately, clearly labeling all required information. Different mistakes should be separated by exactly two line breaks. Each mistake must contain the following information, with each item starting on a new line:\n\nerror_comment: Detailed explanation of the translation mistake.\nerror_span: Exact problematic text from the translation. For omission errors, describe what is missing in brackets, e.g., \"[missing: recent]\"\nerror_severity: Severity rating from 1 (negligible) to 5 (critical).\nerror_type: Semantic or Non-semantic.\nerror_categories: Specific category (choose from mistranslation, undertranslation, overtranslation, omission, grammar, style, do not translate, inconsistency, NE, fluency).\n\nExample entry:\n```\nerror_comment: The translation incorrectly uses 'основной' ('main') for 'actual,' significantly changing the meaning of the original technical term.\nerror_span: основной вход\nerror_severity: 4\nerror_type: semantic\nerror_categories: mistranslation\n```\n\nIf no mistakes are found, output exactly:\n```\nNo translation mistakes found.\n\n## Final scores:\nAccuracy: [Number from 0 to 100]\nFluency: [Number from 0 to 100]\n```\n\nAfter listing all errors, conclude with the final accuracy and fluency scores:\n```\n## Final scores:\nAccuracy: [Number from 0 to 100]\nFluency: [Number from 0 to 100]\n```\n\n---\n### Source Text\n{{source}}\n\n### Translation\n{{translation}}\n"
|
| 6 |
+
}
|
| 7 |
+
]
|
| 8 |
+
}
|
prompts/rate_full_with_examples.md
ADDED
|
@@ -0,0 +1,146 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# RATE full + examples
|
| 2 |
+
|
| 3 |
+
````text
|
| 4 |
+
# Translation Error Analysis Task
|
| 5 |
+
|
| 6 |
+
## Instructions
|
| 7 |
+
You are an expert translator and linguistic reviewer. Your task is to carefully analyze the provided original text and its translation. You must identify and clearly explain any errors or issues according to the guidelines below. You will be provided with the source, translation, and previous context source(optional) and translation. You must evaluate only the given translation, but you should read the context carefully to identify any mistakes related to it.
|
| 8 |
+
|
| 9 |
+
## Guidelines
|
| 10 |
+
|
| 11 |
+
### Error Severity Scale (1 to 5):
|
| 12 |
+
- 4-5 (Major/Critical Errors): These errors significantly distort meaning, lead readers to misunderstand the original message, or are highly inappropriate for the target audience.
|
| 13 |
+
Examples: Incorrect key terminology, significant mistranslation, adding incorrect or misleading information, serious grammatical errors that change meaning, cultural insensitivity/offensiveness, translation that is highly misleading or severely disrupts understanding.
|
| 14 |
+
|
| 15 |
+
- 2-3 (Minor errors): These errors negatively impact readability, fluency, style, or grammar without seriously distorting essential information.
|
| 16 |
+
Examples: Awkward phrasing, unnatural word choices, literal translations, incorrect but minor grammar/punctuation mistakes, inconsistent style or terminology that slightly impacts readability without compromising the main message.
|
| 17 |
+
|
| 18 |
+
- 1 (Negligible or arguable errors): Issues that some translators may interpret differently, debatable stylistic nuances, or very minor mistakes that do not meaningfully affect text quality.
|
| 19 |
+
Example: Using a suitable but less preferable word choice or style variation that is still acceptable in general translation practice.
|
| 20 |
+
|
| 21 |
+
### Types and Categories of Translation Errors
|
| 22 |
+
All errors must be classified into one of two main types: Semantic or Non-semantic. Within each type, select the appropriate specific category.
|
| 23 |
+
|
| 24 |
+
#### Semantic Errors (errors affecting meaning):
|
| 25 |
+
- Mistranslation: Incorrect translation of meaning. This category covers:
|
| 26 |
+
- Significantly incorrect meaning or terminology.
|
| 27 |
+
- Ambiguity that leads to confusion or misunderstanding.
|
| 28 |
+
- Literal translation that misses the intended meaning of idioms/phrases.
|
| 29 |
+
- Errors in concepts critical to understanding the content.
|
| 30 |
+
Examples:
|
| 31 |
+
Source: "Now we have three wires: an actual input, its inverted counterpart and sign of the input."
|
| 32 |
+
Translation: "Теперь у нас есть три провода: основной вход, его инвертированная копия и сигнал на входе."
|
| 33 |
+
Problems here: "основной" ("main") is incorrect for "actual"; "копия" ("copy") is incorrect for "counterpart"; "сигнал на входе" ("input signal") is a mistranslation for "sign of the input."
|
| 34 |
+
|
| 35 |
+
Source: "...digital rights management..."
|
| 36 |
+
Translation: "управление цифровыми правами" is actually "digital rights management," but the better known term in Russian is "защита цифровых прав" or "защита авторских прав на цифровой контент."
|
| 37 |
+
|
| 38 |
+
Source: "...We've all been there."
|
| 39 |
+
Translation: "Мы все были там." ("там" literally means “there” physically, while proper idiomatic meaning would be "Мы все через это проходили.")
|
| 40 |
+
|
| 41 |
+
- Undertranslation: Too general translation; details and important nuances of the original message are lost or diluted significantly.
|
| 42 |
+
Example:
|
| 43 |
+
Source: "Drain the fluid from the FRONT differential."
|
| 44 |
+
Translation: "Слейте жидкость..."
|
| 45 |
+
Problem: Missing critical detail ("differential fluid" not specified).
|
| 46 |
+
|
| 47 |
+
- Overtranslation: Unjustified or overly-specific translation of a general term. Adding unnecessary specific information that was not in the source.
|
| 48 |
+
Example: Translating general "meal" specifically as "обед" ("lunch").
|
| 49 |
+
|
| 50 |
+
- Omission: Deleting words, phrases, or information from the original text without a justified reason. Sometimes, minor omissions with very limited impact may be negligible or minor severity.
|
| 51 |
+
Example:
|
| 52 |
+
Source: "...you need much more than looking up typical interview questions..."
|
| 53 |
+
Translation: "...понадобится больше, чем просто погуглить стандартные вопросы."
|
| 54 |
+
Problem: "much more" should specify "значительно больше" or "заметно больше."
|
| 55 |
+
|
| 56 |
+
#### Non-Semantic Errors (errors that don't directly change meaning):
|
| 57 |
+
- Grammar: Grammar and punctuation errors, spelling mistakes, incorrect capitalization, and typos that don't dramatically affect meaning but lower translation quality and professionalism.
|
| 58 |
+
|
| 59 |
+
- Style: Stylistically awkward or unnatural wording choices. Usually minor but can be critical if style is severely affected.
|
| 60 |
+
Example: Translating "apparel company" as "компания по изготовлению готового платья" (this is stylistically unnatural).
|
| 61 |
+
|
| 62 |
+
- Do Not Translate: Items or information that should remain untranslated (e.g., code snippets, URLs, bibliographic references, names of buttons in software). Undue translation of such elements is considered an error.
|
| 63 |
+
Example: Translating code snippets incorrectly.
|
| 64 |
+
|
| 65 |
+
- Inconsistency: Variations or inconsistent translations within the same text (terminology, names, gender forms). Usually minor but can become major if it causes confusion.
|
| 66 |
+
Example: Translating "squash" inconsistently as both "тыква" and "кабачок" within the same context.
|
| 67 |
+
|
| 68 |
+
- Named Entity (NE): Incorrect translation or transcription of proper nouns (person names, place names, brand/company names). Usually minor in European names if the overall identity is recognizable, major if the name becomes unrecognizable or for languages with complex transliteration rules (e.g., Asian languages). If a standard known translation exists and is ignored, consider severity 4.
|
| 69 |
+
|
| 70 |
+
- Fluency: Issues affecting readability, smoothness, or natural flow of sentences, confusing structure/order of words. This category includes unnatural sentence constructions or very awkward phrasing.
|
| 71 |
+
Example:
|
| 72 |
+
Source: "Rachel Brosnahan to play Lois Lane alongside David Corenswet's Clark Kent."
|
| 73 |
+
Translation: "Рэйчел Броснахэн сыграет Лоис Лэйн вместе с Кларком Кентом в исполнении Дэвида Коренсвета..."
|
| 74 |
+
Problem: This implies Brosnahan will play both Lois Lane and Clark Kent, causing misunderstanding of sentence structure meanings.
|
| 75 |
+
|
| 76 |
+
#### Other
|
| 77 |
+
- Other: Errors that clearly impact the quality or correctness of the translation but do not logically fit into any of the previously defined semantic or non-semantic categories. This category should be used sparingly—only when no other defined categories apply. When using "other," you should explain in detail why the existing categories do not fit, clearly describing the nature and impact of the identified issue.
|
| 78 |
+
|
| 79 |
+
### Final Accuracy and Fluency Assessment
|
| 80 |
+
After identifying all errors, you must provide overall scores for accuracy and fluency on a scale from 0 to 100:
|
| 81 |
+
|
| 82 |
+
Accuracy (0-100): Measures how faithfully the translation conveys the meaning of the source text. Consider whether all information is correctly transferred, whether key concepts are accurately represented, and whether there are any semantic distortions.
|
| 83 |
+
90-100: Near perfect accuracy with minimal or no semantic errors
|
| 84 |
+
70-89: Good accuracy with few minor meaning issues
|
| 85 |
+
50-69: Moderate accuracy with some notable meaning issues
|
| 86 |
+
30-49: Poor accuracy with substantial meaning distortions
|
| 87 |
+
0-29: Very poor accuracy with critical meaning errors or significant missing content
|
| 88 |
+
|
| 89 |
+
Fluency (0-100): Measures how natural, readable, and grammatically correct the translation is in the target language, regardless of accuracy. Consider grammar, word choice, idiomatic expressions, and overall readability.
|
| 90 |
+
|
| 91 |
+
90-100: Reads like original target-language text, perfect grammar and naturalness
|
| 92 |
+
70-89: Generally fluent with minor issues in style or expression
|
| 93 |
+
50-69: Somewhat fluent but with noticeable awkwardness or unnatural phrasing
|
| 94 |
+
30-49: Poor fluency with numerous grammatical errors or unnatural constructions
|
| 95 |
+
0-29: Very poor fluency, difficult to understand, severely ungrammatical
|
| 96 |
+
|
| 97 |
+
### Steps to Analyze the Translation:
|
| 98 |
+
- Read and fully understand the source text.
|
| 99 |
+
- Compare the source text to its translation carefully.
|
| 100 |
+
- If the context source and translation are given, identify any mistakes related to the context. Note that you are to evaluate only the given translation, not the one in the context.
|
| 101 |
+
- Identify any translation errors or issues.
|
| 102 |
+
- Assess overall accuracy and fluency scores.
|
| 103 |
+
- Provide final error report using a clear, simple plain-text list-like format (see "Output Format" section).
|
| 104 |
+
- If there are no translation mistakes identified, your output should be: "No translation mistakes found."
|
| 105 |
+
|
| 106 |
+
### Error Report Format:
|
| 107 |
+
Provide final error report using a clear, simple plain-text list-like format. Each mistake should be described separately, clearly labeling all required information. Different mistakes should be separated by exactly two line breaks. Each mistake must contain the following information, with each item starting on a new line:
|
| 108 |
+
|
| 109 |
+
error_comment: Detailed explanation of the translation mistake.
|
| 110 |
+
error_span: Exact problematic text from the translation. For omission errors, describe what is missing in brackets, e.g., "[missing: recent]"
|
| 111 |
+
error_severity: Severity rating from 1 (negligible) to 5 (critical).
|
| 112 |
+
error_type: Semantic or Non-semantic.
|
| 113 |
+
error_categories: Specific category (choose from mistranslation, undertranslation, overtranslation, omission, grammar, style, do not translate, inconsistency, NE, fluency).
|
| 114 |
+
|
| 115 |
+
Example entry:
|
| 116 |
+
```
|
| 117 |
+
error_comment: The translation incorrectly uses 'основной' ('main') for 'actual,' significantly changing the meaning of the original technical term.
|
| 118 |
+
error_span: основной вход
|
| 119 |
+
error_severity: 4
|
| 120 |
+
error_type: semantic
|
| 121 |
+
error_categories: mistranslation
|
| 122 |
+
```
|
| 123 |
+
|
| 124 |
+
If no mistakes are found, output exactly:
|
| 125 |
+
```
|
| 126 |
+
No translation mistakes found.
|
| 127 |
+
|
| 128 |
+
## Final scores:
|
| 129 |
+
Accuracy: [Number from 0 to 100]
|
| 130 |
+
Fluency: [Number from 0 to 100]
|
| 131 |
+
```
|
| 132 |
+
|
| 133 |
+
After listing all errors, conclude with the final accuracy and fluency scores:
|
| 134 |
+
```
|
| 135 |
+
## Final scores:
|
| 136 |
+
Accuracy: [Number from 0 to 100]
|
| 137 |
+
Fluency: [Number from 0 to 100]
|
| 138 |
+
```
|
| 139 |
+
|
| 140 |
+
---
|
| 141 |
+
### Source Text
|
| 142 |
+
{{source}}
|
| 143 |
+
|
| 144 |
+
### Translation
|
| 145 |
+
{{translation}}
|
| 146 |
+
````
|