Detecting Overlapping Communities Using Distributed Neighbourhood Threshold in Social Networks

Abstract

In this work, we have proposed a simple overlapping community detection algorithm based on a distributed neighbourhood threshold method (DNTM). DNTM uses pre-partitioned disjoint communities and then analyzes the neighbourhood distribution of boundary nodes in disjoint communities to detect overlapping communities. It is a form of seed-based global method since boundary nodes are considered as seeds and become the starting point for detecting overlapping communities. Threshold value for each boundary node is used as minimum influence by the neighbours of a node in order to determine its belongingness to any community. The effectiveness of the DNTM algorithm has been demonstrated by testing on fifteen real-world datasets and compared with seven overlapping community detection algorithms. DNTM outperforms comparable algorithms with 10 out of 15 datasets and gives comparable results for the remaining 5 datasets in terms of the extended modularity Q_{ov} measure. Experiments with various disjoint algorithms on 15 datasets reveal that DNTM with tolerance community detection (TCD) as a preprocessing algorithm gives the best result.