Rational Graph Attention Network and Broad Learning for Emotion Classification in Textual Conversations

Although CBBL was proposed in Chap. 7 and has achieved good results, there is still room for improvement of its performance. Thus, in this chapter, we propose a new method for ECTE task, named RGATBL, which is based on RGAT and BL. RGATBL aims to provide a better solution to capture contextual features and long-distance dependency by using RGAT, RoBERTa, and BL. It is also the first effort to combine RGAT and BL to classify emotion in textual conversations.