KDD Papers

A Location-Sentiment-Aware Recommender System for Both Home-Town and Out-of-Town Mobile Users

Hao Wang (Institute of Software, Chinese Academy of Sciences);Yanmei Fu (Institute of Software, Chinese Academy of Sciences);Qinyong Wang (Institute of Software, Chinese Academy of Science);Changying Du (Institute of Software, Chinese Academy of Sciences);Hongzhi Yin (University of Queensland);Hui Xiong (Rutgers University)


Spatial item recommendation has become an important means to help people discover interesting locations, especially when people pay a visit to unfamiliar regions. Some current researches are focusing on modelling individual and collective geographical preferences for spatial item recommendation based on users’ check-in records, but they fail to explore the phenomenon of user interest drift across geographical regions, i.e., users have different interests when they travel in different regions. Besides, they ignore the sentiment influence of crowds’ reviews for users’ check-in behaviors. Specifically, it is intuitive that users would not check in a spatial item whose overall history reviews seem negative, though it might satisfy their interests. Therefore, it should recommend the item to the user at the location. In this paper, we propose a latent probabilistic generative model called LSARS to mimic the decision-making process of users’ check-in activities both in home-town and out-of-town scenarios by adapting to user interest drift and implicit sentiment for spatial items, which can learn location-aware and sentiment-aware individual interests according to the contents of spatial items and crowds’ reviews. As users’ check-in records left in out-of-town regions are extremely sparse, LSARS incorporates the crowds’ preferences learned from local users’ check-in behaviors. We deploy LSARS to two application scenarios: spatial item recommendation and target user discovery. Extensive experiments on two large-scale location-based social networks (LBSNs) datasets show that LSARS achieves better performance than existing state-of-the-art competing methods.