Friendship prediction based on the fusion of topology and geographical features in LBSN

Friendship prediction in social networks is useful for various applications, such as friend/place recommendation and privacy management. In this paper, we propose a friendship prediction approach by fusing the topology and geographical features in location based social networks (LBSNs). We investigate the features of users' relationship both online and offline and quantify the contributions of selected features through information gain metric. Three key features are selected, namely user social topology, location category, and check-in location. Friendship is predicted based on the fusion of the selected online/offline features. Three inference models are selected to infer the friendship, including Random Forests, Support Vector Machine (SVM), and Naive Bayes. The proposed approach is validated by intensive empirical evaluations using the collected Foursquare and Jiepang datasets.