Generation or Judgement? A Paradigm Perspective on LLM-Based Emotion-Cause Pair Extraction in Conversation

2026-07-29

AI Summary

This work addresses the instability of large language models (LLMs) in dialogue-based emotion-cause pair extraction, where paradigm mismatches hinder their ability to comprehensively generate all valid relation pairs. Through a systematic comparison of generative and discriminative paradigms, the study finds the latter consistently superior. Building on this insight, the authors propose a novel approach that integrates a discriminative framework with a selective re-examination mechanism: candidate pairs are first ranked, and an auxiliary retriever identifies borderline cases to trigger targeted re-evaluation. This design effectively mitigates LLMs’ limitations in relation discovery. Evaluated on three benchmark datasets, the method achieves consistent F1 improvements of 0.50–1.46 points over strong baselines while incurring only 1.49× the inference time, demonstrating a favorable balance between effectiveness and efficiency.

Abstract

Emotion-cause pair extraction in conversation (ECPEC) identifies utterance pairs in which one utterance causes an emotion expressed in another. Recent LLM-based approaches formulate ECPEC at markedly different granularities, ranging from generating complete pair sets to judging individual candidate pairs. In this paper, we make the surprising observation that task formulation substantially affects performance, where pair-level judgement outperforms dialogue-level generation in all 18 controlled comparisons. We investigate the sources of this paradigm gap and find that many relations omitted by dialogue-level generation remain recognizable under explicit pair queries, under which the model recognizes 92.7%-98.1% of emotion-cause relations. This suggests that LLMs can recognize emotion-cause relations but struggle to discover and return complete pair sets. Pair-level judgement alleviates this burden, although its candidate rankings are more reliable than the binary decisions produced by a shared threshold. Based on this diagnosis, we introduce an auxiliary retriever that selectively re-examines ambiguous boundary cases, yielding consistent F1 improvements of 0.50-1.46 points across three datasets while maintaining an inference time of only 1.49x that of the baseline paradigm. These findings show that task decomposition and candidate scope are critical to effectively utilizing LLMs for ECPEC.

Problem

• Emotion-Cause Pair Extraction
• Conversation
• Large Language Models
• Task Formulation
• Pair Judgement

Innovation

• emotion-cause pair extraction
• large language models
• task formulation
• pair-level judgement
• auxiliary retriever