Information theoretic approach for cold start users with diversity improvement technique for semantic recommender system

Abstract

This paper aims to use semantic database for recommendation purposes. It deals with two very specific problems of Recommendation System, namely Cold Start user and Diversity. We first describe cold start users and predict recommendation for them using information theory based methods. To introduce serendipitous results we also include aggregate diversity methods to the predicted ratings. Furthermore, we explain the results obtained from the rated items, and also increase the Intra List Diversity using a ranking-based approach that is different from the popularity-based approach employed in the past.