An Approach to Social Relationship Ranking on Internet-Based Social Platforms by Tempo-spatial Data Mining Using Location Prediction Technique

During the last decade, we have witnessed the prosperity of the Internet-based social platforms and mobile social applications such as Facebook, Twitter, etc. Meanwhile, due to the popularity of mobile terminals such as smart phones and variety of PADs, it is feasible to obtain relatively accurate tempo-spatial data from mobile terminal holders when they visit and upload geo-tagged messages or pictures to Internet based social platform. Therefore, it is observed that the volume of tempo-spatial social data posted on social platforms keeps increasing. This brings us more opportunities to mine the semantic information according to the analysis on the tempo spatial social information collected from Internet-based social platforms. In this paper, we present an approach to social relationship ranking by mining the tempo-spatial social data. To deal with the sparsity of raw tempo-spatial social data, the location prediction technique is employed. According to the comparison between the social relationship ranking method with location prediction and its version without location prediction, it shows that the former outperforms the latter substantially. Finally, we make several helpful observations about the social relationship ranking method when location prediction technique is adopted.