Knowledge Acquisition from Dialogue

Satoshi SATO

Japan Advanced Institute of Science and Technology

Asahidai 15, Tatsunokuchi, Nomi, Ishikawa, 923-12, Japan

e-mail: sato@jaist.ac.jp

We study a model of knowledge acquisition from dialogue. Our final goal is to make a computer program that converse with a human user and acquire information and knowledge from him/her. First of all, we studied a model of understanding of utterances. We believe that interaction between an utterance and a hearer's knowledge plays an important role in the process of understanding of the utterance. From this point of view, we propose a new model that has the following features. 1. Understanding an utterance is the process of matching between the utterance and a hearer's knowledge. Every content word in an utterance has to be connected to a concept in a hearer's knowledge. If every content word is connected to a concept, the utterance is understood and the new information that was carried by the utterance is acquired. 2. The response to the utterance is determined by the result of the matching. If the connecting process of a word X ends in failure, the response "What is a X?" is produced.

Keywords: Dialogue, Understanding, Knowledge Acquisition, Situation of Information