Sink Node Elimination to Enhance the Performance of Overlapping Detection Algorithms along with Comparison of Existing Algorithm

Abstract

Objectives: To eliminate Sink nodes so that rate of detection can improve within the community overlapping detection and this also increases the modularity. It also consumes less time. Methods/Statistical Analysis: Modified k-clique Algorithm is used. Clique algorithm considers Sink nodes. Sink nodes are those which do not have any connecting edge. The proposed algorithm (MKC) does eliminate these nodes and hence consider only those nodes which are connected in nature. To detect the Sink nodes adjacency matrix is used. MATLAB is used for the simulation. Community detection toolbox is used which provides several functions for graph generations, clustering algorithms etc., Findings: This approach produces better result in terms of community finding. More community are discovered with the proposed approach and also entropy is improved greatly. Result in terms of time consumption is reduced almost by 50%. Application/Improvements: The length and complexity of the cliques found is reduced considerably. The speed is almost enhanced by 5% which can further be increased by using hop count mechanism in addition to the already used Sink node elimination.