This repository contains the official implementation of our paper ECERC: Evidence-Cause Attention Network for Multi-Modal Emotion Recognition in Conversation, published at ACL 2025.

The experiments were conducted on a Windows 10 operating system equipped with an NVIDIA A100 GPU (80GB). Further system specifications are provided in the accompanying OS_info.txt and requirement.yml files.

conda env create -f requirement.yml -n ecerc

The benchmark datasets used in our paper are IEMOCAP and MELD. Due to copyright restrictions, we provide links to the preprocessed versions only. The original datasets can be downloaded from their respective official sources.

You can download our pretrained ECERC_MODEL(" "/"_2"/"_3") for each dataset from our Huggingface Repository: zt-ai/ECERC. After downloading, place the models in the corresponding IEMOCAP/MELD folder. To reproduce results closely matching those reported in our paper (which presents the average over 3 runs), you can execute the following commands:

cd IEMOCAP/MELD
python inference.py

You can also run the train.py script to train the model from scratch. For accurate reproduction of our results, please ensure that your experimental environment matches ours exactly, as specified in 1. Requirements.

@inproceedings{zhang-tan-2025-ecerc,
    title = "{ECERC}: Evidence-Cause Attention Network for Multi-Modal Emotion Recognition in Conversation",
    author = "Zhang, Tao  and
      Tan, Zhenhua",
    editor = "Che, Wanxiang  and
      Nabende, Joyce  and
      Shutova, Ekaterina  and
      Pilehvar, Mohammad Taher",
    booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2025.acl-long.102/",
    pages = "2064--2077",
    ISBN = "979-8-89176-251-0"
}

This code repository is licensed under the MIT License. ECERC supports commercial use.