Identification of Influential Spreaders in Social Networks using Improved Hybrid Rank Method

Abstract

Abstract Online Social Networks provide us a gateway towards valuable information regarding personal preferences, interests, and connections. Ranking nodes by centrality measures help us to identify some of the most critical and influential nodes in a network who play a crucial role in the information spreading process and influence maximization. Traditional centrality measures have certain restraints and do not provide optimal results single-handedly. With the advances in research, leading to the formulation of the concept of hybrid centrality, the detection of the most influential spreaders in a network has shown a large scale of improvement. Some of the real-world phenomena such as viral marketing, rumor spreading, etc. can be addressed by leveraging the fact that only the dominant and authoritative nodes play a crucial role in propagating information about a new product or idea and those nodes can also help in blocking false information or rumors from reaching the other parts of the network. In this paper, we present the Improved Hybrid Rank algorithm, which combines two centralities, namely, the Extended Neighborhood Coreness centrality and the H-Index centrality. The results obtained by simulating our proposed method using the SIR (Susceptible-Infected-Recovered) model on different un-directed and directed real-world networks show that the ranking of nodes and the choice of influential spreaders, as proposed by our algorithm outperforms many other classical methods.