Memetic algorithm based location and topic aware recommender system

Abstract

Recommender systems based on locations and tags have received a great deal of interest over the last few years. Whereas, recent advances do not transcend limits of recommendation algorithms that solely use geographical information or textual information. In this paper, we propose a novel location and tag aware recommendation framework that uses ratings, locations and tags to generate recommendation. In this framework, all users are partitioned into several clusters by a newly designed Memetic Algorithm (MA) based clustering method. Normal users are recommended items obtained by applying Latent Dirichlet Allocation (LDA) to users within each cluster. For cold-start users, each cluster is viewed as a new user. Each cluster is recommended a list of items by applying LDA to all clusters. The recommendation list to the querying cluster is recommended to all cold-start users in this cluster. Extensive experiments on real-world datasets demonstrate that compared with state-of-the-art location and tag aware recommendation algorithms, the proposed algorithm has better performance on making recommendations and alleviating cold-start problem.