Jaccard Based Similarity Index in Graphs: A Multi-Hop Approach

Prediction of connections between nodes in graphs has gained widespread attention in the recent years as more and more data is shifted to non-euclidean format. While local similarity metrics like Jaccard, Adamic Adar, Resource Allocation have been widely used until now, they fail to convey the similarity measure when two nodes are separated by more than one hops. In this paper we have proposed a similarity index that is aligned in the lines of Jaccard Coefficient but manages to predict the similarity between two nodes fundamentally on the basis of the shortest path between them.