FeatureMF: An Item Feature Enriched Matrix Factorization Model for Item Recommendation

Matrix Factorization (MF) is one of the most successful Collaborative Filtering (CF) techniques used in recommender systems due to its effectiveness and ability deal with very large user-item rating matrix. However, when matrix sparseness increases performance deteriorates. Expanding MF include side-information users items has been shown by many researchers both improve general recommendation help alleviate data-sparsity cold-start issues CF. In regard item feature side-information, schemes incorporate this information through a two stage process: intermediate results (e.g., on similarity) are first computed based attributes; these then combined MF. paper, focussing we propose model that directly incorporates features into framework single step process. The model, which name FeatureMF, does projecting every available attribute datum each same latent factor space items, thereby effect enriching representation Results presented comparative experiments against three state-of-the-art enriched models, as well four reference benchmark using public real-world datasets, Douban Yelp, training:test ratio scenarios each. It yield best over all models across contexts including situations, particular, achieving 0.9% 6.5% MAE improvement next HERec. FeatureMF also found cold start scale well, almost linearly, computational time, function dataset size.