CBLA: A Clique Based Louvain Algorithm for Detecting Overlapping Community

Abstract

Graph mining is one of the significant tasks in the field of computer science. Most of the applications generate a vast
amount of data which is represented with the help of a graph. Due to this graph representation, these applications have
become complex and increased in size. Finding relevant information from that graph becomes a complicated task. For
this community detection algorithms play a vital role in graph partitioning to retrieve relevant information. Finding
communities in a graph reduces the complexity of the graph due to related data comes closer to forming a
community. Many algorithms have been introduced in the last decade; the Clique percolation method (CPM) is the
benchmark algorithm for finding an overlapping community. But in this method, some nodes remain unclassified,
nodes that are not part of the clique. Paper proposed the clique-based Louvain algorithm(CBLA), which can classify the
non-classified node (NCN) obtained after finding cliques in one of the communities by applying the Louvain algorithm.
Louvain algorithm is used to classify the non-overlapped community, but with the help of cliques, it will also detect the
overlapped nodes. This paper compared the proposed algorithm with four other benchmark algorithms. The proposed
algorithm gives equal or enhanced performance among all compared algorithms.