An approach for hesitant node classification in overlapping community detection

Overlapping community detection has recently drawn much attention in the field of social network analysis. In this paper, we propose a notion of hesitant node (HN) in network with overlapping community structure. An HN is a special kind of node that contacts with multiple communities but the communication is not frequent or even accidental, thus its community structure is implicit and its classification is ambiguous. Besides, HNs are not rare to be found in networks and may even take up a large number of the nodes in the network, just like the long tail. They should either be classified into certain communities which would promote their development in the network or regarded as the hubs if they are the efficient junctions between different communities. Current approaches have difficulties in identifying and processing HNs. In this paper, a quantitative method based on the Density-Based Rough Set Model (DBRSM) is proposed by combining the advantages of density-based algorithms and rough set model. Our experiments on the real-world and synthetic datasets show the advancement of our approach. HNs are classified into communities which are more similar with them and the classification process enhances the modularity as well.