Location-Aware Service Recommendation With Enhanced Probabilistic Matrix Factorization

Owing to the ever-growing popularity of mobile computing, a large number of services have been developed for a variety of users. Considering this, recommending useful services to users is an urgent problem that needs to be addressed. Collaborative filtering (CF) approaches have been successfully adopted for services recommendation. Nevertheless, the prediction accuracy of the existing CF approaches is likely to reduce due to many reasons, such as inability to use side information and high data sparsity, which further lead to low quality of services recommendation. In order to solve these problems, some model-based CF approaches have been proposed. In this paper, we propose a novel quality of service prediction approach based on probabilistic matrix factorization (PMF), which has the capability of incorporating network location (an important factor in mobile computing) and implicit associations among users and services. First, we propose a novel clustering method that is capable of utilizing network location to cluster users. Based on the clustering results, we further propose an enhanced PMF model. The proposed model also incorporates the implicit associations among users and services. In addition, our model incorporates the implicit relationships between the users and the services. We conducted experiments on one real-world data set, and the experimental results show that our model outperforms the compared methods.