An Efficient Partition-Based Approach to Identify and Scatter Multiple Relevant Spreaders in Complex Networks.

Area of Science:

• Graph theory
• Network analysis
• Computational social science

Background:

• Identifying influential nodes (spreaders) is vital for understanding and controlling diffusion processes in networks.
• Existing methods often focus on node centrality rankings but neglect their spatial distribution within the network topology.
• Network topology significantly impacts the dynamics of spreading phenomena.

Purpose of the Study:

• To propose a novel method for identifying relevant and well-scattered spreaders in graph analysis.
• To develop a strategy for optimally determining the number of spreaders to select.
• To enhance the efficiency and control of spreading processes by considering both node relevance and distribution.

Main Methods:

• A two-phase approach involving graph partitioning and subsequent identification/distribution of relevant nodes.
• Utilizing the underlying graph topology to ensure selected nodes are both influential and geographically dispersed.
• Applying the SIR (Susceptible-Infected-Recovered) spreading model to evaluate the proposed method on real-world complex networks.

Main Results:

• The proposed method identified more influential and scattered spreaders compared to established algorithms like degree, closeness, Betweenness, VoteRank, HybridRank, and IKS.
• Combining the distribution strategy with classical metrics (e.g., degree) improved propagation influence.
• The approach demonstrated superior performance over computationally intensive strategies.

Conclusions:

• The proposed method effectively identifies influential and well-distributed spreaders, optimizing network analysis for spreading processes.
• It offers a computationally efficient solution applicable to large-scale networks.
• The findings suggest that considering node distribution alongside centrality is key for effective network control and information flow.