Method of topic processing for cooperative dialog systems
Hidemitsu Furukawa, N. Harada, Kôichi Yamada, A. Nukuzuma, K. Ishimaru · 2002
Discourse information such as context and topics must be tracked in order to achieve cooperative interaction in a decision support system with natural language interfaces. This paper presents a method for processing topics, one aspects of the context, in goal-oriented dialogs. In our method, topics are recognized by matching in a fuzzy thesaurus and tracked by using a topic stack. Results of application to an information consultation system show our method can effectively recognize topics from surface expressions.>