Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Do LLMs Feel? Teaching Emotion Recognition with Prompts, Retrieval, and Curriculum Learning

Xinran Li, Yu Liu, Jiaqi Qiao, Xiujuan Xu

TL;DR

This work tackles Emotion Recognition in Conversation (ERC) by integrating explicit and implicit emotion interpretations into prompts, a dedicated demonstration retrieval repository, and a curriculum-based training regime for large language models. The PRC-Emo framework leverages a retrieval-template module and LoRA-finetuned LLMs to retrieve relevant demonstrations and external knowledge during inference, organized by an easy-to-hard curriculum based on weighted emotional shifts. Empirical results on IEMOCAP and MELD demonstrate state-of-the-art performance, with ablations confirming the value of each component, particularly the prompt design and cross-speaker curriculum. The approach advances ERC by combining prompt engineering, retrieval-augmented reasoning, and progressive learning, enhancing robustness and generalization across diverse conversational domains.

Abstract

Emotion Recognition in Conversation (ERC) is a crucial task for understanding human emotions and enabling natural human-computer interaction. Although Large Language Models (LLMs) have recently shown great potential in this field, their ability to capture the intrinsic connections between explicit and implicit emotions remains limited. We propose a novel ERC training framework, PRC-Emo, which integrates Prompt engineering, demonstration Retrieval, and Curriculum learning, with the goal of exploring whether LLMs can effectively perceive emotions in conversational contexts. Specifically, we design emotion-sensitive prompt templates based on both explicit and implicit emotional cues to better guide the model in understanding the speaker's psychological states. We construct the first dedicated demonstration retrieval repository for ERC, which includes training samples from widely used datasets, as well as high-quality dialogue examples generated by LLMs and manually verified. Moreover, we introduce a curriculum learning strategy into the LoRA fine-tuning process, incorporating weighted emotional shifts between same-speaker and different-speaker utterances to assign difficulty levels to dialogue samples, which are then organized in an easy-to-hard training sequence. Experimental results on two benchmark datasets -- IEMOCAP and MELD -- show that our method achieves new state-of-the-art (SOTA) performance, demonstrating the effectiveness and generalizability of our approach in improving LLM-based emotional understanding.

Do LLMs Feel? Teaching Emotion Recognition with Prompts, Retrieval, and Curriculum Learning

TL;DR

This work tackles Emotion Recognition in Conversation (ERC) by integrating explicit and implicit emotion interpretations into prompts, a dedicated demonstration retrieval repository, and a curriculum-based training regime for large language models. The PRC-Emo framework leverages a retrieval-template module and LoRA-finetuned LLMs to retrieve relevant demonstrations and external knowledge during inference, organized by an easy-to-hard curriculum based on weighted emotional shifts. Empirical results on IEMOCAP and MELD demonstrate state-of-the-art performance, with ablations confirming the value of each component, particularly the prompt design and cross-speaker curriculum. The approach advances ERC by combining prompt engineering, retrieval-augmented reasoning, and progressive learning, enhancing robustness and generalization across diverse conversational domains.

Abstract

Emotion Recognition in Conversation (ERC) is a crucial task for understanding human emotions and enabling natural human-computer interaction. Although Large Language Models (LLMs) have recently shown great potential in this field, their ability to capture the intrinsic connections between explicit and implicit emotions remains limited. We propose a novel ERC training framework, PRC-Emo, which integrates Prompt engineering, demonstration Retrieval, and Curriculum learning, with the goal of exploring whether LLMs can effectively perceive emotions in conversational contexts. Specifically, we design emotion-sensitive prompt templates based on both explicit and implicit emotional cues to better guide the model in understanding the speaker's psychological states. We construct the first dedicated demonstration retrieval repository for ERC, which includes training samples from widely used datasets, as well as high-quality dialogue examples generated by LLMs and manually verified. Moreover, we introduce a curriculum learning strategy into the LoRA fine-tuning process, incorporating weighted emotional shifts between same-speaker and different-speaker utterances to assign difficulty levels to dialogue samples, which are then organized in an easy-to-hard training sequence. Experimental results on two benchmark datasets -- IEMOCAP and MELD -- show that our method achieves new state-of-the-art (SOTA) performance, demonstrating the effectiveness and generalizability of our approach in improving LLM-based emotional understanding.

Paper Structure

This paper contains 27 sections, 5 equations, 2 figures, 9 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Traditional Methods
  4. LLM-based Methods
  5. Curriculum Learning
  6. Methodology
  7. Problem Definition
  8. Extraction of External Supplementary Knowledge
  9. Retrieval Template Module
  10. Components of Retrieval Template Module
  11. Demonstration Retrieval Repository
  12. Curriculum Learning
  13. Difficulty Measure Function
  14. Training Scheduler
  15. Experimental Settings
  16. ...and 12 more sections

Figures (2)

  • Figure 1: An example of Explicit Emotion Recognition in Conversation (EERC) and Implicit Emotion Recognition in Conversation (IERC). In this conversation, although the girl feels apprehension due to the boy’s anxious utterance, “I wonder if we can make it,” she expresses optimism in her own response to encourage both of them.
  • Figure 2: PRC-Emo’s architecture has two main stages: extracting external supplementary knowledge and predicting emotion labels, with curriculum learning applied during training. The two prompts at the bottom-left extract explicit and implicit emotion interpretations and speaker characteristics as external knowledge. This information is passed to the bottom-right prompt, which performs the final emotion recognition by retrieving similar pairs from a retrieval repository to aid the process.