Neu-PCM: Neural-based potential correlation mining for POI recommendation

In the past few years, with the advent and developments of artificial intelligence, the location-based mobile services became prevalent, which produce large-scale location-based data that embeds abundant hints of user preferences on locations. Point of interest (POI) recommendation, as one of the significant mobile services, aims to recommend new satisfactory location to user according to their historical records. Existing classical POI recommendation models based on matrix factorization and collaborative filtering both face a significant challenge that they can not capture the users’ preferences deeply and effectively. Hence, we propose a deep POI recommendation model (Neu-PCM), which is based on neural networks, to extract the potential correlation between user and location. First, we present a local learning and dimension-reduction network to obtain the key information. Second, we put forward a union network to mine the potential correlation effectively. Third, we build a deep matrix factorization to enhance the final prediction by correlation combination. The results of experiments on real-world datasets demonstrate our model outperforms the current popular recommendation algorithms.