An Effective Conversational Agent with User Modeling Based on Bayesian Network

A conversational agent is an agent that interacts with users using natural language interface. Especially in Internet space, its role has been recently highlighted as a virtual representative of a web site. However, most conversational agents use simple pattern matching techniques for their answer without considering user goals and thus give user an unsatisfactory response. In this paper, we propose a conversational agent that utilizes user model constructed on Bayesian network for its responses consistent with user goals. The agent is applied to the active guide of a website, which shows that the user modeling based on Bayesian network helps respond to user queries appropriately and consistently with the user goals.