Collective Factorization for Relational Data: An Evaluation on the Yelp Datasets

Matrix factorization has found incredible success and widespread application as a collaborative filter-ing based approach to recommendations. Unfortu-nately, incorporating additional sources of incom-plete and noisy evidence is quite difficult to achieve in such models, however this information is often crucial for obtaining further gains in accuracy. For example, in the Yelp datasets, additional information about businesses from reviews, categories, and at-tributes should be leveraged for predicting ratings, even though these are often inaccurate and partially-observed. Instead of creating customized solutions that are specific to the types of evidences, in this pa-per we present a generic approach to factorization of relational data that collectively models all the re-lations in the database. By learning a set of factors that are shared across all the relations, the model is able to incorporate observed information from all the relations, while also predicting all the relations of interest. Our evaluation on four Yelp datasets demon-strates effective utilization of additional information for held-out user preference and attribute prediction, but further, we present accurate models even for cold-start businesses for which we do not observe any rat-ings or attributes. We also present joint visualization