Spark’s GraphX-based link prediction for social communication using triangle counting

Link prediction in a given instance of a network topology is a crucial task for extracting and inspecting the evolution of social networks. It predicts missing links in existing community networks and new or terminating links in future systems. It also attracted much attention in many fields. In the past decade, many methodologies have been compiled to predict the suitable links in a given social network. Analyzing link prediction methods is difficult when the network is very complex due to restrictive computing cost. It is still a very challenging task to predict missing links efficiently and accurately in an incomplete complex network. Depending on the certainty, the nodes with an incredible number of normal neighbors will probably be connected. Numerous similarity indices have accomplished extensive exactness and efficiency that greatly optimized this task. To accommodate this instance, in this paper, we propose one such index, namely Clustering Coefficient Index, using triangle counting implemented on the component of Apache Spark’s GraphX methodology. The proposed index uses the property of formation of triangles in the given network topology and clustering coefficients. Experimental results show that the proposed methodology outperforms in linking the suitable communications compared to other existing methods.