Mining Business Opportunities from Location-based Social Networks

摘要

Urbanization’s rapid progress has modernized a large number ofrnhuman beings’ lives. This urbanization progress is accompaniedrnby the increase of a variety of shops (e.g., restaurants and fitnessrncenters) to meet the increasing citizens, which means businessrnopportunities for the investors. Nevertheless, it is difficult for therninvestors to catch such opportunities because opening what kind ofrnbusiness at which place is not easy to decide. In this paper, we takernthis challenge and define the business opportunity mining problem,rnwhich recommends new business categories at a partitioned businessrndistrict. Specifically, we exploit the data from location-basedrnsocial networks (LBSNs) to mine the business opportunities, guidingrnthe business owners to open new commercial shops in certainrncategories at a particular area. First, we define the properties ofrna business district and propose a greedy algorithm to partition arncity into different districts. Next, we propose an embedding modelrnto learn latent representations of categories, which captures thernfunctional correlations among business categories. Furthermore, wernpropose a ranking model based on the pairwise loss to recommendrncategories for a specific district. Finally, we conduct experiments onrnYelp data, and experimental results show that our proposed methodrnoutperforms the baseline methods and resolves the problem well