Enhancing a Location-based Recommendation System by Enrichment with Structured Data from the Web

Location-based social networks (LBS) enable users to checkin at points of interests (POIs), share this information with other users within the network, and receive recommendations about new and interesting POIs in their vicinity. In this paper, we show how such recommendations can be improved by adding background information from Linked Open Data and other sources of structured data. Using a dataset previously crawled from the LBS Gowalla, we analyze which types of background information are the most beneficial for improving the recommendations. In a series of offline experiments, we show that the quality of recommendations can be improved by 51% in precision and 150% in recall.