Emotion-Enhanced Multi-Task Learning with LLMs for Aspect Category Sentiment Analysis

TL;DR

The paper tackles the limitation of aspect category sentiment analysis (ACSA) that focuses on coarse polarity by introducing an emotion-enhanced multi-task learning framework. It leverages large language models to generate category-level emotions grounded in Ekman's six emotions and refines them via a Valence-Arousal-Dominance (VAD) mapping, creating high-quality emotion supervision. The model jointly learns category-sentiment and category-emotion predictions with a weighted loss, improving fine-grained affective reasoning and generalization. Empirical results on SemEval Restaurant and Laptop datasets show consistent improvements over classical ACSA baselines and instruction-based fine-tuning approaches, validating the effectiveness of integrating affective dimensions into ACSA for more nuanced sentiment analysis.

Abstract

Aspect category sentiment analysis (ACSA) has achieved remarkable progress with large language models (LLMs), yet existing approaches primarily emphasize sentiment polarity while overlooking the underlying emotional dimensions that shape sentiment expressions. This limitation hinders the model's ability to capture fine-grained affective signals toward specific aspect categories. To address this limitation, we introduce a novel emotion-enhanced multi-task ACSA framework that jointly learns sentiment polarity and category-specific emotions grounded in Ekman's six basic emotions. Leveraging the generative capabilities of LLMs, our approach enables the model to produce emotional descriptions for each aspect category, thereby enriching sentiment representations with affective expressions. Furthermore, to ensure the accuracy and consistency of the generated emotions, we introduce an emotion refinement mechanism based on the Valence-Arousal-Dominance (VAD) dimensional framework. Specifically, emotions predicted by the LLM are projected onto a VAD space, and those inconsistent with their corresponding VAD coordinates are re-annotated using a structured LLM-based refinement strategy. Experimental results demonstrate that our approach significantly outperforms strong baselines on all benchmark datasets. This underlines the effectiveness of integrating affective dimensions into ACSA.

Figures (4)

• Figure 1: Overview of our emotion-enhanced multi-task ACSA framework.
• Figure 2: An example of using an LLM to decompose sentences.
• Figure 3: Prompt template employed for the emotion generation.
• Figure 4: Prompt template employed for the revised emotion generation.