Identifying influential nodes in Social Networks: Neighborhood Coreness based voting approach
Efficient modeling of information diffusion in an online social network, like viral distribution of a market product or rumor control, can be achieved through the most influential nodes in the system. Hence, to pass the information to a maximum extent of the network or keep it confined to a lesser e...
Saved in:
Published in | Physica A Vol. 553; p. 124215 |
---|---|
Main Authors | , |
Format | Journal Article |
Language | English |
Published |
Elsevier B.V
01.09.2020
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | Efficient modeling of information diffusion in an online social network, like viral distribution of a market product or rumor control, can be achieved through the most influential nodes in the system. Hence, to pass the information to a maximum extent of the network or keep it confined to a lesser extent in the case of rumor, it is essential to find the influential nodes. Many classical centralities have been proposed in literature with certain limitations. Recently Vote Rank based method was introduced to find the seed nodes. It selects a set of spreaders based on a voting scheme where voting ability of each node is same and each node gets the vote from its neighbors. But we argue that the voting ability of each node should be different and should depend on its topological position in the network. In this paper, we propose a coreness based VoteRank method called NCVoteRank to find spreaders by taking the coreness value of neighbors into consideration for the voting. Experiments and simulations using Susceptible–Infected–Recovered (SIR) stochastic model on many real datasets show that our proposed method, NCVoteRank, outperforms some of the existing popular methods such as PageRank, K-shell, Extended Coreness, VoteRank, and WVoteRank.
•Based on the notion of coreness, NCVoteRank is proposed for influence maximization.•It identifies spreaders from the diverse positions that lead to improved results.•Simulation on various benchmark datasets suggest improved performance of the proposed method. |
---|---|
ISSN: | 0378-4371 1873-2119 |
DOI: | 10.1016/j.physa.2020.124215 |