M3GAT: A Multi-modal, Multi-task Interactive Graph Attention Network for Conversational Sentiment Analysis and Emotion Recognition.

Autor: YAZHOU ZHANG, AO JIA, BO WANG, PENG ZHANG, DONGMING ZHAO, PU LI, YUEXIAN HOU, XIAOJIA JIN, DAWEI SONG, JING QIN
Zdroj: ACM Transactions on Information Systems; Jan2024, Vol. 42 Issue 1, p1-32, 32p
Abstrakt: The article focuses on a "Multi-Modal, Multi-Task Interactive Graph Attention Network" (M3GAT) designed for conversational sentiment analysis and emotion recognition. Topics include understanding human sentiments and emotions in multi-modal conversations, addressing the interdependence of sentiment and emotion, and incorporating conversational context dependency.
Databáze: Complementary Index