Hybrid Recommender System Using Latent Features

This paper proposes a hybrid recommender system that utilizes latent features. The main problem discussed in this paper is the cold start problem. To handle this problem, the proposed system first extracts latent features from items represented by a multi-attributed record using a probabilistic model. Then, it calculates the similarity of users from their ratings. Both similarities between items and users are used for predicting unknown rating of a user to a item. We evaluate the proposed method using a movie data set and shows that the proposed method achieves good performance for small ratings information.