A Recommendation-type Dialogue System Responding to Potential Requests in Consideration of Personal Attributes

Abstract

A dialogue system that responds mechanically to given queries can be used for public relations for prospective university students. However, as a user-specific problem, collecting information from first-time entrances and passive examinations is difficult, and a conventional dialogue system may not collect sufficient information. Therefore, we developed a recommendation-type dialogue system that supports information gathering by evoking potential user requests. Based on the preference analysis results using previous and real-time query histories and recommended information from universities, information in the form of queries is recommended to users when considering their personal attributes. We evaluated the presence or absence of a recommendation function and implemented an introduction evaluation targeting public relations for those who wish to enroll in a university. The average number of queries increases by evoking potential requests, and the information gathering of the user is supported.