Aggregated Temporal Tensor Factorization Model for Point-of-Interest Recommendation

Point-of-interest (POI) recommendation is an important application in location-based social networks (LBSNs), which mines user check-in sequences to suggest interesting locations for users. Because user check-in behavior exhibits strong temporal patterns—for instance, users would like to check-in at restaurants at noon and visit bars at night. Hence, capturing the temporal influence is necessary to ensure the high performance in a POI recommendation system. Previous studies observe that the temporal characteristics of user mobility in LBSNs can be summarized in three aspects: periodicity, consecutiveness, and non-uniformness. However, previous work does not model the three characteristics together. More importantly, we observe that the temporal characteristics exist at different time scales, which cannot be modeled in prior work. In this paper, we propose an Aggregated Temporal Tensor Factorization (ATTF) model for POI recommendation to capture the three temporal features together, as well as at different time scales. Specifically, we employ a temporal tensor factorization method to model the check-in activity, subsuming the three temporal features together. Next, we exploit a linear combination operator to aggregate temporal latent features’ contributions at different time scales. Experiments on two real-world data sets show that the ATTF model achieves better performance than the state-of-the-art temporal models for POI recommendation.