Towards Diversified Local Users Identification Using Location Based Social Networks

Chao Huang, Dong Wang, Shenglong Zhu

Abstract: Identifying a set of diversified users who are local residents in a city is an important task for a wide spectrum of applications such as target ads of local business, surveys and interviews, and personalized recommendations. While many previous studies have investigated the problem of identifying the local users in a given area using online social network information (e.g., geotagged posts), few methods have been developed to solve the diversified user identification problem. In this paper, we propose a new analytical framework, Diversified Local Users Finder (DLUF), to accurately identify a set of diversified local users using a principled approach. In particular, the DLUF scheme first defines a new distance metric that measures the diversity between local users from physical dimension. The DLUF scheme then provides a solution to find the set of local users with maximum diversity. The performance of DLUF scheme is compared to several representative baselines using two real world datasets obtained from Foursquare application. We observe that the DLUF scheme accurately identifies the local users with a great diversity and significantly outperforms the compared baselines.