Learning classifier system for generating various types of dialogues in conversational agent

Abstract

Most of the conversational agents respond to the users in an unsatisfactory way because of using the simple sequential pattern matching. In this paper, we propose a conversational agent that can respond with various sentences for improving the user's familiarity. The agent identifies the user's intention using DA (Dialogue Acts) and increases the intelligence and the variety of the conversation using LCS (Learning Classifier System). We apply this agent to the introduction of a web site. The results show that the conversational agent has the ability to present more adequate and friendly response to user's query. © Springer-Verlag 2003.