Personalized recommendation system for Sina MicroComic users

The MicroComic Website underneath the Chinese biggest social network Sina.com (Comic.sina.com.cn) provides a broad range of comics for a huge set of users. The demand of a personalized recommendation system for comic readers has been put on the table recently to improve the users' experience. This paper first investigates the classic item-based collaborative filtering and latent factor model via utilizing the official data sets provided by the Sina.com Company. Then a hybrid recommendation approach is proposed to enhance the recommender performance. The recommendation system can estimate each individual user's preference of the unread comics and generate recommendation list of comics with relative high ratings. The proposed recommender achieves not only good prediction accuracy but outperforms the current recommendation system in respect of Coverage.