Emotion-Cause Pair Extraction in Conversations via Semantic Decoupling and Graph Alignment

TL;DR

This paper introduces SCALE, a novel framework for emotion-cause pair extraction in conversations that employs semantic decoupling and graph alignment to improve accuracy and capture complex causal relations.

Contribution

The paper proposes a new semantic decoupling and global alignment approach for ECPEC, achieving state-of-the-art results and better modeling conversational causality.

Findings

SCALE outperforms existing methods on benchmark datasets.

Semantic decoupling improves the distinction between emotion and cause semantics.

Optimal transport enables many-to-many emotion-cause matching.

Abstract

Emotion-Cause Pair Extraction in Conversations (ECPEC) aims to identify the set of causal relations between emotion utterances and their triggering causes within a dialogue. Most existing approaches formulate ECPEC as an independent pairwise classification task, overlooking the distinct semantics of emotion diffusion and cause explanation, and failing to capture globally consistent many-to-many conversational causality. To address these limitations, we revisit ECPEC from a semantic perspective and seek to disentangle emotion-oriented semantics from cause-oriented semantics, mapping them into two complementary representation spaces to better capture their distinct conversational roles. Building on this semantic decoupling, we naturally formulate ECPEC as a global alignment problem between the emotion-side and cause-side representations, and employ optimal transport to enable many-to-many and globally consistent emotion-cause matching. Based on this perspective, we propose a unified framework SCALE that instantiates the above semantic decoupling and alignment principle within a shared conversational structure. Extensive experiments on several benchmark datasets demonstrate that SCALE consistently achieves state-of-the-art performance. Our codes are released at https://github.com/CoCoSphere/SCALE.