Construction of Automatic Utterance Model based on Non-linearity of Natural Language for Realizing Natural Conversation between Humans and Computers

本研究基于语言的非线性特征，建立了一个自动化的话语模型。特别是提出了主题的选取方法和关联系统，为主题的选取提供了依据。我们使用Twitter的日志数据和使用Wikipedia关于主题选择的层次知识库启用了约90%的准确率的主题选择。对概念库的自动构造方法进行了评价。

In this research, we constructed a automatically utterance model based on the non-linearity of lingua. Especially, we proposed the selection method of topics and the association system that provided the basis of topics selection. We enabled the topic selection with about 90% accuracy using Twitter’s log data and the hierarchy knowledge base using Wikipedia about topics selection. We estimated the automatically construction method of Concept-base.