license: cc-by-nc-sa-4.0
language:
- en
tags:
- youtube,
- review,
- sentiment analysis,
- emotion recognition,
- unimodality and multimodality
task_categories:
- text-classification
- audio-classification
- image-classification
size_categories:
- n<1K
20251212
This work is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International.
20251208
For academic purposes only.
The sentiment annotations are ready to share. Other annotations will follow soon.
These are the features used in the paper LDW: Label Divergence Weighting for Multimodal Sentiment Analysis.
Contact [email protected] for the access.
UniC: a Dataset for Emotion Analysis of Videos with Multimodal and Unimodal Labels
UniC is comprised of 965 emotion-rich video clips selected from YouTube, annotated in text, audio, (silent) video and multimodal setups with both categorical and dimensional labels.
Categorical label: disgust, disappointment, neutral, confusion, surprise, contentment, and joy.
Dimensional label: Valence and arousal.
If you use this dataset, please cite our papers:
-- Quanqi Du, Sofie Labat, Thomas Demeester and Veronique Hoste. UniC: a dataset for emotion analysis of videos with multimodal and unimodal labels. Language Resources & Evaluation 59, 2857–2892 (2025). https://doi.org/10.1007/s10579-025-09837-0
-- Quanqi Du, Loic De Langhe, Els Lefever, and Veronique Hoste. 2025. LDW: Label Divergence Weighting for Multimodal Sentiment Analysis. In Proceedings of the 33rd ACM International Conference on Multimedia (MM '25). Association for Computing Machinery, New York, NY, USA, 12342–12351. https://doi.org/10.1145/3746027.3758160
Contact: [email protected]