贾熹滨, 李让, 胡长建, 陈军成. 智能对话系统研究综述[J]. 北京工业大学学报, 2017, 43(9): 1344-1356. DOI: 10.11936/bjutxb2016090023
    引用本文: 贾熹滨, 李让, 胡长建, 陈军成. 智能对话系统研究综述[J]. 北京工业大学学报, 2017, 43(9): 1344-1356. DOI: 10.11936/bjutxb2016090023
    JIA Xibin, LI Rang, HU Changjian, CHEN Juncheng. Review of Intelligent Dialogue System[J]. Journal of Beijing University of Technology, 2017, 43(9): 1344-1356. DOI: 10.11936/bjutxb2016090023
    Citation: JIA Xibin, LI Rang, HU Changjian, CHEN Juncheng. Review of Intelligent Dialogue System[J]. Journal of Beijing University of Technology, 2017, 43(9): 1344-1356. DOI: 10.11936/bjutxb2016090023

    智能对话系统研究综述

    Review of Intelligent Dialogue System

    • 摘要: 智能对话系统作为人工智能领域的核心技术,即将成为新的和谐人机交互方式,具有重大的研究意义和应用价值,因此,较为全面、深入地总结了深度学习模型在对话生成中的应用及对话系统领域的研究进展和现状.首先,阐述了智能对话系统的类型划分,介绍系统的模块框架构成,包括各模块的主要研究问题与关键技术的主流思路和研究现状;然后,从理论模型、研究进展及可用性等角度深度剖析了现有的对话生成技术解决方案,重点分析了应用于自然语言生成的序列到序列的神经网络结构及搭建原理;最后,对存在的问题进行总结,并展望了未来的研究方向.

       

      Abstract: Intelligent dialogue system is the core technology in the field of artificial intelligence, and it will act as a new harmonious human-computer interaction. The research has great theory significance and application value. The status and advances of dialogue system and the application of deep learning model in dialogue generation were comprehensively summarized in this paper.First, the type of intelligent dialogue system was elaborated. The module framework of the system was introduced, including the research contents of the modules and key components of the main technology ideas and their research status. Then, the various solutions to dialogue generation were systematically analyzed from the perspective of theoretical models, advances and usability. The neural network structures and the construction principle of sequence-to-sequence applied to natural language generation were analyzed emphatically. Finally, the problems and prospects for future research directions were summarized.

       

    /

    返回文章
    返回