Exploring celebrities on inferring user geolocation in twitter

Abstract

Location information of social media users provides crucial context to monitor real-time events such as natural disasters, terrorism and epidemics. Since only a small amount of social media data are geotagged, inference techniques play a substantial role to predict user spatial locations by incorporating characteristics of their behavior. Based on utilized source of information, related works are divided into text-based (based on text posted by users), network-based (based on the friendship network), and some hybrid methods. In this paper, we propose a novel approach based on the notion of celebrities to infer the location of Twitter users. We categorize highly-mentioned users (celebrities) into local and global, and consequently utilize local celebrities as a major location indicator for inference. A label propagation algorithm is then utilized over a refined social network for geolocation inference. Finally, we propose a hybrid approach by merging a text-based method as a back-off strategy into our network-based approach. Empirical experiments using three standard Twitter benchmark datasets demonstrate the superior performance of our approach over the state-of-the-art methods.