计算机应用研究2024，Vol.41Issue(1)：1-9,9.DOI:10.19734/j.issn.1001-3695.2023.05.0206
共情对话研究进展
Research progress of empathetic dialogue system
摘要
Abstract
Artificial intelligence-driven dialogue is a current research hotspot with a wide range of promising applications.However,such dialogue systems currently generally lack emotional interaction capabilities,limiting their application in psy-chological care,depression and other mental disorders.How to make dialogue systems fully understand users'emotions and generate responses with empathy is one of the main challenges facing dialogue systems today.This paper first introduced two major challenges in empathic dialogue research:emotion perception and empathic dialogue generation,and summarised the relevant research methods in separate studies.Emotion perception tasks could be broadly classified into rule-based,machine-learning and deep-learning approaches,while empathic dialogue generation could be broadly classified into retrieval-based and dynamic generation-based approaches.Then this paper introduced the latest developments in empathic dialogue,and sum-marised the features and links of empathic dialogue datasets,generic dialogue datasets and multimodal datasets,and sum-marised different evaluation methods in current empathic dialogue research to facilitate subsequent research.Finally,this pa-per presented a summary and outlook of the research work on empathic dialogue.关键词
共情对话/情感感知/对话生成Key words
empathetic dialogue/emotional perception/dialogue generation分类
信息技术与安全科学引用本文复制引用
何俊,饶方喜,周志豪,徐秋..共情对话研究进展[J].计算机应用研究,2024,41(1):1-9,9.基金项目
国家自然科学基金资助项目(62066025) （62066025）
