Capturing Missing Links in Social Networks Using Vertex Similarity

The vertex similarity measure is a useful tool to discover and justify the relationship of vertices in a complex network. We propose the Relation Strength Similarity (RSS), a new vertex similarity measure that utilizes the network topology to discover similar vertices. Compared to other vertex similarity measures, RSS has the following advantages. First, it is an asymmetric metric which allows the measure to be used in more general social network applications. Second, it can be employed on a weighted network, in which the relation strength of two neighboring nodes can be explicitly expressed. Third, users could adjust the “discovery range” parameter for better performance based on their domain knowledge. Using coauthorship network as experimental data, our method outperforms other vertex similarity measures in terms of the ability to predict future coauthoring behavior among scholars.