LBSNRank: personalized pagerank on location-based social networks

Different from traditional social networks, the location-based social networks allow people to share their locations according to location-tagged user-generated contents, such as checkins, trajectories, text, photos, etc. In location-based social networks, which are based on users' checkins, people could share his or her location according to checkin while visiting around. However, people's locations change frequently and the rankings of people change dynamically too, which makes ranking on graphs a challenging work. To address this challenge, we propose the LBSNRank algorithm on graphs with nodes whose contents change dynamically. To validate our algorithm on real datasets, we have crawled and analyzed a dataset from the Dianping website. Experiments on this real dataset show that our LBSNRank algorithm performs better than traditional personalized PageRank in efficiency.