A Book Recommendation System Considering Sentiment of Content of Interest

Abstract

Although the benefits of reading are widely recognized, many people do not read despite their interest. To find books that reflect their interests using previous book recommendations, users need keywords or a browsing history related to such books. Accordingly, it is difficult to obtain satisfactory results for users who do not already have good reading habits. In this study, we propose a book recommendation system that enables not only users with solid reading habits but also users who rarely read to easily get results that reflect their own interests. The proposed method uses the user’s contents of interest as a query and determines recommended books based on similarities of vectors of the contents and related emotions, contained in tweets about the user’s contents of interest and book reviews. In this study, we confirmed the effectiveness of the proposed method through experiments in which users expressed their impressions in a questionnaire.