Pair-Based Joint Learning with Relational Graph Convolutional Networks for Emotion-Cause Pair Extraction

Anonymous

Abstract: Emotion-cause pair extraction (ECPE) aims to extract the emotion clauses and the corresponding cause clauses, which have recently received more attention. Previous methods sequentially encode features with a specified order, which first encode the emotion and cause features for clause extraction and then combine them for pair extraction, leading to an imbalance in inter-task feature interaction where features extracted later have no direct contact with the former. To this end, we propose a novel joint encoding network, which generates pairs and clauses features simultaneously in a joint feature learning manner to model the causal relationship from clauses. Specifically, from a multi-relational perspective, we construct a heterogeneous undirected graph and apply the Relational Graph Convolutional Network (RGCN) to capture the complex relationship between clauses and the relationship between pairs and clauses. Experimental results show that our model achieves state-of-the-art performance on the Chinese benchmark corpus.