Accuracy Optimization of Centrality Score Based Community Detection

Abstract

Various concepts can be represented as a graph or the network. The network representation helps to characterize the varied relations between a set of objects by taking each object as a vertex and the interaction between them as an edge. Different systems can be modelled and analyzed in terms of graph theory. Community structure is a property that seems to be common to many networks. The division of the some objects into groups within which the connections or relations are dense, and the connections with other objects are sparser. Various research and data points proves that many real world networks has these communities or groups or the modules that are sub graphs with more edges connecting the vertices of the same group and comparatively fewer links joining the outside vertices. The groups or the communities exhibit the topological relations between the elements of the underlying system and the functional entities. The proposed approach is to exploit the global as well as local information about the network topologies. The authors propose a hybrid strategy to use the edge centrality property of the edges to find out the communities and use local moving heuristic to increase the modularity index of those communities. Such communities can be relevantly efficient and accurate to some applications.
DOI: 10.17762/ijritcc2321-8169.150737