CauAIN: Causal Aware Interaction Network for Emotion Recognition in Conversations

• DOI: 10.24963/ijcai.2022/628
• Publication Date: 2022-07-16T02:55:56Z
• Authors: Weixiang Zhao; Yanyan Zhao; Xin Lu
• Keywords: Utterance; Natural language understanding

Emotion Recognition in Conversations has attained increasing interest the natural language processing community. Many neural-network based approaches endeavor to solve challenge of emotional dynamics conversations and gain appealing results. However, these works are limited capturing deep clues conversational context because they ignore emotion cause that could be viewed as stimulus target emotion. In this work, we propose Causal Aware Interaction Network (CauAIN) thoroughly understand with help detection. Specifically, retrieve causal provided by commonsense knowledge guide process utterance traceback. Both traceback steps performed from perspective intra- inter-speaker interaction simultaneously. Experimental results on three benchmark datasets show our model achieves better performance over most baseline models.