EmpKI: Empathetic Dialogue Generation Model Integrating Common Sense Knowledge and Emotional Intent
Year: 2024
Type: conference-paper
Abstract: As the dialogue systems continue to evolve, enhancing their emotional intelligence becomes increasingly crucial, thus empathetic dialogue generation tasks become more significant. Current empathetic dialogue generation models primarily focus on the emotional aspects of users, generating responses aligned with those emotions. However, they often overlook the emotional intent of the users, potentially resulting in emotionally correct but contextually divergent responses, consequently diminishing user experience and weakening user engagement. To elevate the quality of empathetic dialogue generation, this paper proposes a method that integrates common sense knowledge and emotional intent. This method utilizes a common sense knowledge reasoning module to deduce relevant information and an emotional intent prediction module to anticipate user intent. It concurrently considers user intent and common sense knowledge, selecting appropriate common sense information based on the user's intent. Experiments conducted on the Empathetic Dialogues dataset demonstrate a substantial enhancement in both the content and emotional expression of the generated responses.
Authors: Ting Meng, Xiaoxia Bi, Shuai Liu
Institution: Xi’an University of Posts and Telecommunications
Language: English
Topic: Speech and dialogue systems
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4411730086