Mining Influential Spreaders in Complex Networks by an Effective Combination of the Degree and K-Shell

TL;DR

This paper proposes a new method combining degree and k-shell indices to more accurately and efficiently identify influential nodes in complex networks, outperforming previous models.

Contribution

It introduces an effective hybrid approach that improves the accuracy and reduces the computational complexity of influential spreader detection in networks.

Findings

Outperforms previous models in accuracy and efficiency

Demonstrates superior results in Kendall's Tau and monotonicity index

Shows better correlation and lower time complexity

Abstract

Graph mining is an important technique that used in many applications such as predicting and understanding behaviors and information dissemination within networks. One crucial aspect of graph mining is the identification and ranking of influential nodes, which has applications in various fields including marketing, social communications, and disease control. However, existing models and methods come with high computational complexity and may not accurately distinguish and identify influential nodes. This paper develops a method based on the k-shell index and degree centrality of nodes and their neighbors. Comparisons to previous works, such as Degree and Neighborhood information Centrality (DNC) and Neighborhood and Path Information Centrality (NPIC), are conducted. The evaluations, which include the correctness with Kendall's Tau, resolution with monotonicity index, correlation plots, and time complexity, demonstrate its superior results.