Personalized Commodity Recommendations of Retail Business using User Feature based Collaborative Filtering

Collaborative filtering is an extensively adopted approach for commodity recommendation. This paper proposes a user feature based collaborative filtering algorithm named UFCF for personalized commodity recommendations of retail business. It adopts matrix factorization and user features that are extracted from users' behaviors to improve the accuracy of recommendation result and alleviate the impact of sparse data. Experiments with real datasets from a supermarket marketing group demonstrate the effectiveness of the algorithm.