Context-Aware Location Annotation on Mobility Records Through User Grouping

Due to the increasing popularity of location-based services, a massive volume of human mobility records have been generated. At the same time, the growing spatial context data provides us rich semantic information. Associating the mobility records with relevant surrounding contexts, known as the location annotation, enables us to understand the semantics of the mobility records and helps further tasks like advertising. However, the location annotation problem is challenging due to the ambiguity of contexts and the sparsity of personal data. To solve this problem, we propose a Context-Aware location annotation method through User Grouping (CAUG) to annotate locations with venues. This method leverages user grouping and venue categories to alleviate the data sparsity issue and annotates locations according to multi-view information (spatial, temporal and contextual) of multiple granularities. Through extensive experiments on a real-world dataset, we demonstrate that our method significantly outperforms other baseline methods.