OSUSUME: cross-lingual recommender system for research papers

In this paper, we introduce a cross-lingual recommender system for research papers based on multiple-facets in Japanese. The system is the first Japanese research paper recommender system and recommends international papers simply by typing Japanese keywords. Academic search engine users vary from undergraduate students to researchers. As the users have different backgrounds and preferences, they will aim at different kind of papers. In order to consider the various users' preferences, the system introduces eight viewpoints: internationality, similarity, state of the art, serendipity, contextual analysis (target and method), essentiality and authority. The recommendation results of our system were evaluated by a questionnaire survey. Based on the survey results, we discuss the relevance of the proposed viewpoints and the precision of the system.