Location Traceability of Social Media Analysis in Urban Complex

Online social media (e.g., Weibo, Twitter and Facebook) has emerged as a novel paradigm of sensing collecting observations (commonly called claims) from humans about the physical world in this big data era. These observations may contain location information directly based location-based sensor network (LBSN) or not, and maybe part of them, we can predict the information of location by inference through text, image or other metadata. Hence a challenging problem in social sensing lies in whether we can gain the location information. This problem is referred as location traceability problem. In Recent years, urban complex has been more and more popular and played an important role in the smart city. The location traceability problem, if well addressed, directly contributes to help police and property company to have an effective management in urban complex and improve their public image. However, two main challenges exist in the current location traceability solutions. The first one is the lack of location information without the help of LBSN scenario, where a great deal of sources are disseminating the information which is lack of location information, making the location traceability problem difficult. The second challenge is the large range of location information in urban complex scenario. However, the location information we need is the indoor location in the urban complex. In this paper, we developed a Heuristic Location Traceability (HLT) scheme to handle the above two challenges. Especially, the HLT scheme considers the claims itself that humans may report their location information in the texts, images or other metadata and develop a novel algorithm to predict the location information with the help of these metadata. Experimental results from a real world dataset show that our HLT scheme significantly outperforms other location traceability methods.