Extended core-based community detection for directed networks

The focus of this paper is on detecting overlapping communities for the directed graphs by implementing a new algorithm and analyzing it with various performance metrics. The algorithm aims at finding core nodes for the directed graph which are subset of communities and have higher contact frequency. These are then extended to find communities using compactness measurement (CM). The compactness of a node to the community is defined as the ratio of the outward degree of the node to the community to that of the total out degree of that node. Another approach that will be used to extend communities around core nodes is based on similarity measurement (SM) - two nodes are said to be similar if they share more mutual neighbours. We are able to achieve a success rate of 70% when CM is used and about 10–15% with SM based expansion method. The proposed algorithm is also compared with the existing method for community detection.