管道式对话系统研究进展及其在医疗领域应用  

Advances in Pipelined Dialogue System Research and Its Application in Medical Field

作　　者：杜建强[1,2] 郑奇民 罗计根 聂斌[1] 熊旺平[1] 刘勇 周添强 DU Jian-qiang;ZHENG Qi-min;LUO Ji-gen;NIE Bin;XIONG Wang-ping;LIU Yong;ZHOU Tian-qiang(College of Computer Science,Jiangxi University of Traditional Chinese Medicine,Nanchang 330004,China;Key Laboratory of Artificial Intelligence in Chinese Medicine,Jiangxi University of Traditional Chinese Medicine,Nanchang 330004,China)

机构地区：[1]江西中医药大学计算机学院,南昌330004 [2]江西中医药大学中医人工智能重点研究室,南昌330004

出　　处：《科学技术与工程》2024年第6期2187-2200,共14页Science Technology and Engineering

基　　金：国家自然科学基金(82260988,82160955,82274680);江西中医药大学校级科技创新团队发展计划(CXTD22015)。

摘　　要：随着人工智能技术的快速发展,任务型对话系统成为人机交互领域的热点研究方向。管道式方法是其一种经典的设计框架,在任务型对话系统的研究和应用中扮演着重要角色。对管道式任务型对话系统的研究进展进行了综述,并重点探讨了其在医疗领域的应用。首先介绍管道式对话系统各模块的基本原理、评价指标以及常用的数据集。然后,梳理了近年来深度学习技术在管道式对话系统研究中取得的重要进展,并进一步归纳了所用模型的优缺点。接着,重点关注了管道式对话系统在医疗领域的应用,并讨论了医疗对话系统的需求和挑战。最后,总结并展望了未来的研究方向和发展趋势。With the rapid development of artificial intelligence technology,task-based dialogue systems have become a hot research direction in the field of human-computer interaction.The pipelined approach is regarded as a classical design framework,playing a significant role in the research and application of task-based dialogue systems.The research progress in pipelined task-based dialogue systems was overviewed,with a specific emphasis on their application in the medical field.Firstly,the basic principles,evaluation metrics and commonly used datasets of each module of pipelined dialogue systems were introduced.Then,the important advances made by deep learning techniques in the research of pipelined dialogue systems in recent years were sorted out,and the advantages and disadvantages of the models used were further summarized.Subsequently,the application of pipelined dialogue systems in the medical field was specifically focused on,and the requirements and challenges of medical dialogue systems were discussed.Finally,the full paper was summarized and future research directions and trends were outlined.

关 键 词：自然语言处理 任务型对话系统 管道式 医疗应用 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]