A Book Recommendation Method Based on Paragraph Vector and User's Book Arrangement
Tatusya Miyamoto and Daisuke Kitayama
Lecture Notes in Engineering and Computer Science: Proceedings of The International MultiConference of Engineers and Computer Scientists 2019,pp.247-251,Hong Kong,2019年3月,査読有
論文PDF
概要
In recent years, the evaluation of recommender systems has focused on not only accuracy but aspects such as novelty, diversity, explainability, coverage, and serendipity as well. This is because users are not always satisfied with book recommender systems focusing on accuracy alone. Recent evaluation measures provide a solution for this problem. In this paper, we propose a book recommendation method that considers serendipity and explainability based on user book arrangement. We focus on the book arrangement such that if the user arranges the books based on his/her own preferences, we perceive that even the same set of books will be arranged differently by each user. For example, a user arrenges for increasing comedy atmosphere, on the other hand, other user arrenges for increasing difficulty of mystery. We propose a book recommender system that understands the intention of a user’s book arrangement and recommends books accordingly. We investigate the effectiveness of our proposed approach through experiments on novels from book review sites.