[3Rin4] Interactive 1

[3Rin4-71] Personalization of Extractive Summarization for Conversational News Contents Delivery

〇Hiroaki Takatsu¹, Mayu Okuda¹, Yoichi Matsuyama¹, Hiroshi Honda², Shinya Fujie^1,3, Tetsunori Kobayashi¹ (1.Waseda University, 2.Honda R&D Co.,Ltd., 3.Chiba Institute of Technology)

Keywords:Automatic Text Summarization, Personalization, Spoken Dialogue System

We are developing a spoken dialogue system that efficiently delivers a massive amount of information like news articles. Here, "efficient" means that only the necessary information is delivered except unnecessary information for the user from target articles. In this system, any given written documents, such as news articles, can be translated into an utterance plan consisting of a primary plan for delivering main content and the associated subsidiary plans for supplementing the main content. A primary plan is automatically generated by applying text summarization techniques. However, in the conventional method, summaries are generated based only on the importance of contents. Therefore, we propose a method of generating personalized summaries for each user by using user's profile that can be obtained from a questionnaire conducted at the start of use. A questionnaire survey was conducted for Waseda University students, who were asked whether they would like to be informed of each sentence in the news articles, as well as questions about their profile, such as interest of genres. The results showed that summaries generated based on interest level estimated using user's profile can transmit information more efficiently than summaries generated based only on the importance of contents.