Paper: UniMSE: Towards Unified Multimodal Sentiment Analysis and Emotion Recognition EMNLP(2022)
Arxiv: https://arxiv.org/pdf/2211.11256.pdf
Abstract:
Multimodal sentiment analysis (MSA) and emotion recognition in conversation (ERC) are key research topics for computers to understand human behaviors. From a psychological perspective, emotions are the expression of affect or feelings during a short period, while sentiments are formed and held for a longer period. However, most existing works study sentiment and emotion separately and do not fully exploit the complementary knowledge behind the two. In this paper, we propose a multimodal sentiment knowledge-sharing framework (UniMSE) that unifies MSA and ERC tasks from features, labels, and models. We perform modality fusion at the syntactic and semantic levels and introduce contrastive learning between modalities and samples to better capture the difference and consistency between sentiments and emotions. Experiments on four public benchmark datasets, MOSI, MOSEI, MELD, and IEMOCAP, demonstrate the effectiveness of the proposed method and achieve consistent improvements compared with the state-of-art methods.
Architecture:
Code: we will open the source codes in the future, please wait patiently.
Features:
BaiDu disk
Link:https://pan.baidu.com/s/190tw6g0bPkiOPu5xPpvDTQ
pwd:xyg5
Link:https://pan.baidu.com/s/17n_Hi2Tv0a7qsPoOp6qLKw pwd:rms6
Link:https://pan.baidu.com/s/1VOgH_BW08TxVEQto8fUvfw pwd:txvm
Text Information Link:https://pan.baidu.com/s/11hRiiTculJTVtsOuMhSAFQ pwd:de1p
Google drive iemocap: https://drive.google.com/file/d/10XPUZo8qb1ILF5_Z17AAODbrKX3_h4wL/view?usp=sharing meld:https://drive.google.com/file/d/1pWH2xPVZFymxeJUrd6gF37qYbvmhh32s/view?usp=sharing mosei:https://drive.google.com/file/d/1NmlzPqOGiGaYbzfRi_12r2BjmKfGP3Hw/view?usp=sharing mosi:https://drive.google.com/file/d/1alEn_5RfHFXuu8jwEFMIbMOLF-28--k_/view?usp=sharing text:https://drive.google.com/file/d/1F4K75-_aj29Qyd1xaA19p-LLA0hbI9g0/view?usp=sharing
- datasets
MOS -- MOSI + MOSEI MOSELD -- MOSI+MOSEI+MELD MOSELDMP -- MOSI+MOSEI+MELD+IEMOCAP
- Environment
torch 1.7.0+pai
torchvision 0.8.0
tensorboardX 2.5
tensorflow-estimator 1.15.1
tensorflow-gpu 1.15.0
transformers 4.12.5
Running
First, generate Universal Label for each dataset -> Simcse folder, versation 3
Second, generate Train dataset based on single dataset MOSI, MOSEI, IEMOCAP,MELD -> preprocess.py & create_dataset.py
Third, setting the dataset folder for dataloader. ->main.py
卑微求引用
@inproceedings{hu-etal-2022-unimse, title = "{U}ni{MSE}: Towards Unified Multimodal Sentiment Analysis and Emotion Recognition", author = "Hu, Guimin and Lin, Ting-En and Zhao, Yi and Lu, Guangming and Wu, Yuchuan and Li, Yongbin", booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing", month = dec, year = "2022", address = "Abu Dhabi, United Arab Emirates", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2022.emnlp-main.534", pages = "7837--7851" }