Spatiotemporal recommendation with big geo-social networking data

Recommendation has become an important mobile application on location-based social networks (LBSNs), especially when users travel to a new place far away from their home. Compared to traditional recommender systems, this type of recommendation is very challenging. A user on geo-social network usually visits only a very limited number of spatial items (points of interest), resulting in sparse user-item matrix. As most users tend to visit the spatial items nearby their homes, the user-item matrix will become even sparser when users travel to a distant place. Another major challenge is that, users’ interests and behavior patterns tend to vary dramatically across different time period and different geographical regions. In this chapter, we focus on effective spatial item recommendation by exploiting both spatial and temporal information on geo-social networks. To solve the sighted challenges, we propose ST-SAGE, a spatial- temporal sparse additive generative (SAGE) model for spatial item recommendation. ST-SAGE considers both personal interests of the users and the preferences of the crowd in the target region at the given time by exploiting both the co-occurrence patterns of spatial items and the content of spatial items. To further alleviate the data sparsity issue, ST-SAGE exploits the geographical correlation by smoothing the crowd’s preferences over a well-designed spatial index structure called spatialpyramid. To speed up the training process of ST-SAGE, we implement a parallel version of the model inference algorithm on the GraphLab framework. We conduct extensive experiments, and the experimental results clearly demonstrate that ST-SAGE outperforms the state-of-the-art recommender systems in terms of recommendation effectiveness, model training efficiency and online recommendation efficiency.