APP下载

主动联想型对话机器人的设计与实现

2021-04-06于丹温涛闫晓宇孙伟来关军

软件工程 2021年2期

于丹 温涛 闫晓宇 孙伟 来关军

摘  要:人机对话系统通常仅是用户命令的执行者,很少能主动发起对话,或依据对话以外的内容展开个性化的人机交互。本文提出了联想型对话机器人的概念和对应设计方案,即通过融合用户提问和用户个人特征的方式产生对话内容,这样的对话系统不仅可以主动发出对话邀请,也可以个性化地回答用户问题,提升人机对话的满意度。文章首先介绍了联想型对话机器人与其他类型对话系统的区别与联系,探讨了联想型对话机器人设计所需的关键技术和方法,并以在线教育系统为例,基于有限状态机的方法,构建了主动联想型对话机器人算法模型和联想式对话的生成策略,使机器人能够联想用户的历史行为,生成以用户使用产品为背景的智能对话。

关键词:联想型机器人;对话系统;有限状态机

Abstract: The human-machine dialogue system is usually only the executor of the user's command, and rarely initiates a dialogue that proactively develops personalized human-machine interaction based on content outside but associative to the dialogue. This paper proposes concept and corresponding design of the associative chatbot, that is, dialogue content is generated by integrating user questions and user personal characteristics. Such a dialogue system can not only actively send out dialogue invitations, but also answer user questions in a personalized way, and improve satisfaction of human-machine dialogue. The paper first introduces differences and connections between the associative chatbot and other types of dialogue systems, and discusses key technologies and methods required for designing an associative chatbot. Then, based on Finite-state Machine (FSM) method, this paper takes online education system as an example and constructs algorithm model of proactively associative chatbots and generation strategies of associative dialogues. Therefore, chatbots can associate user's historical behavior with new conversations and generate intelligent dialogues based on user's situation.

Keywords: associative chatbot; dialogue system; Finite-state Machine

1   引言(Introduction)

當用户初次使用多服务人机交互平台[1]系统时,通常由于不熟悉系统而产生迷茫感,需要在使用前阅读庞杂的使用教程。随着用户使用进程的推进,理想的情况是多个子服务之间可以产生联动,平台可以主动地为他们推送个性化定制的服务,例如自动提示用户经常访问的链接即将失效,根据其经常使用的功能做用户行为分析,并加入与用户的互动中等。但遗憾的是,此类功能很少出现在目前已上线的学习平台系统中。

人机对话系统,也有人将其称为对话机器人,已经被广泛应用于人机交互过程等中的个性化沟通或帮助用户完成特定的任务[2,3]。……

登录APP查看全文