Comparative Analysis of Classic Clustering Algorithms and Girvan-Newman Algorithm for Finding Communities in Social Networks

Nowadays finding patterns in large social network datasets is a growing challenge and an important subject of interest. One of current problems in this field is identifying clusters within social networks with large number of nodes. Social network clusters are not necessarily disjoint sets; rather they may overlap and have common nodes, in which case it is more appropriate to designate them as communities. Although many clustering algorithms handle small datasets well, they are usually extremely inefficient on large datasets. This paper shows comparative analysis of frequently used classic graph clustering algorithms and well-known Girvan-Newman algorithm that is used for identification of communities in graphs, which is especially optimized for large datasets. The goal of the paper is to show which of the algorithms give best performances on given dataset. The paper presents real problem of data clustering, algorithms that can be used for its solution, methodology of analysis, results that were achieved and conclusions that were derived.