The emergence of location-based social networks provides an unprecedented chance to study the interaction between human mobility and social relations. This work is a step towards quantifying whether a location is suitable for conducting social activities, and the notion is named location sociality. Being able to quantify location sociality creates practical opportunities such as urban planning and location recommendation. To quantify a location’s sociality, we propose a mixture model of HITS and PageRank on a heterogeneous network linking users and locations. By exploiting millions of check-in data generated by Instagram users in New York and Los Angeles, we investigate the relation between location sociality and several location properties, including location categories, rating and popularity. We further perform two case studies, i.e., friendship prediction and location recommendation, experimental results demonstrate the usefulness of our quantification.
ACM Conference on Hypertext and Social Media