APTNESS: Incorporating Appraisal Theory and Emotion Support Strategies for Empathetic Response Generation

TL;DR

This work tackles empathetic response generation by distinguishing cognitive and affective empathy and introducing APTNESS, a retrieval-augmented framework that leverages an Appraisal Theory–driven Empathetic Database (APT) and emotional support strategies. It constructs a comprehensive empathetic palette (7 main categories, 23 subcategories) and decomposes emotions into emotion, factors, and situation to guide responses, building a large external resource of 9,663 dialogues. A two-stage generation process retrieves semantically similar responses from APT and predicts appropriate emotional-support strategies via a fine-tuned LoRA, enabling richer, more nuanced empathetic outputs. Evaluations on ED and ET with multiple foundation models show that APTNESS improves both cognitive and affective empathy over baselines, highlighting the value of combining retrieval augmentation with strategy-guided guidance for empathetic dialogue systems. The work provides a practical pathway to more emotionally intelligent AI assistants, supported by publicly released code and datasets.

Abstract

Empathetic response generation is designed to comprehend the emotions of others and select the most appropriate strategies to assist them in resolving emotional challenges. Empathy can be categorized into cognitive empathy and affective empathy. The former pertains to the ability to understand and discern the emotional issues and situations of others, while the latter involves the capacity to provide comfort. To enhance one's empathetic abilities, it is essential to develop both these aspects. Therefore, we develop an innovative framework that combines retrieval augmentation and emotional support strategy integration. Our framework starts with the introduction of a comprehensive emotional palette for empathy. We then apply appraisal theory to decompose this palette and create a database of empathetic responses. This database serves as an external resource and enhances the LLM's empathy by integrating semantic retrieval mechanisms. Moreover, our framework places a strong emphasis on the proper articulation of response strategies. By incorporating emotional support strategies, we aim to enrich the model's capabilities in both cognitive and affective empathy, leading to a more nuanced and comprehensive empathetic response. Finally, we extract datasets ED and ET from the empathetic dialogue dataset \textsc{EmpatheticDialogues} and ExTES based on dialogue length. Experiments demonstrate that our framework can enhance the empathy ability of LLMs from both cognitive and affective empathy perspectives. Our code is released at https://github.com/CAS-SIAT-XinHai/APTNESS.

Figures (6)

• Figure 1: An example of empathetic response from ExTES dataset. Some empathetic responses are based on cognitive empathy, while others based on affective empathy.
• Figure 2: The structure of the APTNESS framework is comprised of three core components: the generation of a empathetic response database with empathetic response appraisal theory, the retrieval augmentation module, and the integration of emotional support strategies module.
• Figure 3: Empathetic Emotional Palette
• Figure 4: Prompt for strategy integration LoRA SFT.
• Figure 5: Prompt for the two-stage empathetic response generation