Discovering associations in heterogeneous social networks

Abstract

The studying of association between behavioral patterns in online social network and influences outside the network is an emerging topic in social network analysis. Our work attempted to study whether it is possible to transfer the analysis model from popular social network sites to those sites which are less popular. If such transference is feasible, the existing social network analyses can be efficiently spread to numerous small and medium social networks in the world. A set of abstract data models, which are named as Generic Networks Data Models (GNDMs), and one abstract association model, which is named as Generic Network Data Association (GNDA), are proposed in our work. The GNDMs conceptually solve the differences in contents among the heterogeneous social networks, while the GNDA defines an abstract model on the associations between online and offline social networks. The GNDA has two components. The first one is borrowed from the studies of popular online social network, while the second one is defined according to the features of the online social network. Therefore, our work is called "transplantation of association analysis" because the association analysis is contributed from one online social network and is used in other networks or applications. A service-based analytical framework (the D-Miner Service Framework) is proposed to implement the GNDA by integrating all relevant solutions proposed in our work. The framework uses a novel linking technique, which is called Generic Network Data Linking (GNDL), to connect the data in a form of the GNDMs. Networks Content Linkage (NWCL), which is based on the GNDL, is developed to automatically connect the news media with the online social network.