“人机对话-聊天机器人”与话语修辞

"Man-machine Dialog-chatbot"and Discourse Rhetoric

本文首先通过"图灵测试"与"中文屋"思想实验两个案例,说明人工智能与语言运用及其表达效果关系紧密。接着介绍ELIZA与PARRY这两个"聊天机器人"的前辈系统,说明这种人机对话的成功依赖于语言交际的社会性与互动合作的主动性。然后介绍目前智能对话系统的基本构架与模块(NLU-DM-NLG),特别展示在研制这种智能对话系统的过程中涉及到的修辞问题,包括话题结构的组织、语言表达的多样性、语言运用的个人风格等。最后以智能对话系统的构建为镜子观照语言学研究,讨论语言运用的概率性与统计学特征,说明语义分析对于语言理解及相关智能系统开发的重要性。

This paper probes into the roles that discourse rhetoric plays in human-machine conversation.First,two famous cases,"Turing Test"and"Chinese Room"thinking experiments,are used to explain the close relation between artificial intelligence and the expressiveness of linguistic performance.Secondly,two pioneer systems of chatbots,that is,ELIZA and PARRY,are used to show that the success of man-machine dialogues depends upon the sociality of linguistic communications and the subjectivity of interactive coorperations.Thirdly,three major modules that current intelligent systems for question answering have are introduced,that is,natural language understanding(NLU),dialogue management(DM),and natural language generation(NLG).A number of rhetoric issues in these systems are addressed,including organization of topic structures,diversity of linguistic expressions,and individual styles of linguistic performance.Finally,the construction of intelligent dialogue systems is used as a mirror to shed light on linguistic research,to show the probabilistic and statistic features of linguistic performance,and to illustrate the importance of semantic analysis in language understanding and in the development of relevant intelligent interaction systems.