A Novel Approach for Analyzing the Social Network
Saranya Balaguru, Rachel Nallathamby, C. R. Rene Robin · Procedia Computer Science · 2015
Massive datasets are becoming more prevalent. In this paper, we propose an algorithm to process a large symmetric matrix of billion scale graph in order to extract knowledge from graph dataset. For example, interesting patterns like the people who frequently visit your page and the most number of participating triangles can be obtained using the algorithm. These interesting patterns are discovered by computation of several eigen values and eigen vectors. The main challenge in analyzing the graph data are simplifying the graph, counting the triangles, finding trusses. These challenges are addressed in the proposed algorithm by using orthogonalization, parallelization and blocking techniques. The proposed algorithm is able to run on highly scalable MapReduce environment. we use a social network dataset (facebook approximately 2 to 7 TB of data) to evaluate the algorithm. we also show experimental results to prove that the proposed algorithm scale well and efficiently process the billion scale graph.