Nearness and influence based link prediction (NILP) in distributed platform

Abstract

Link prediction is a trending research direction in the field of social network analysis due to its vast application especially in the field of network evolution analysis like discovering missing links, identifying positive and negative links etc. It is also used for recommending commodities in e-commerce sites and suggesting friends in online social network. The objective of this paper is to predict hidden or missing links in a directed or undirected, unweighted social network using distributed platform like Spark, which is found to be both accurate and reasonably robust due to its scalable nature. This approach is found to be effective as compared to conventional methods as it considers number of parameters together such as the influence of a node, community structure of the network and the shortest paths between nodes. This approach is found to be efficient as compared to other path-based approaches as it has comparatively less time complexity and preferable over other node-based approaches owing to its accuracy with reasonable computational time.