A machine learning based node influence analysis using isolating and betweenness centrality measures
Quantifying node influence in complex networks is essential for exploring diffusion dynamics and identifying key spreaders. Traditional centrality measures, which rely primarily on network topology, often fail to accurately capture information propagation and neglect the interplay between local and global structural properties. To address these limitations, this paper proposes a machine learning-based framework incorporating a novel Isolating-Betweenness Centrality (ISBC) measure that integrates both local and global network information. The proposed ISBC combines local and global topological features to enable a more accurate and comprehensive evaluation of the influence of nodes. Furthermore, the framework incorporates the infection rate as a crucial feature to better reflect the diffusion behavior in the real-world. The proposed approach formulates node influence estimation as a classification problem. Feature vectors are constructed using conventional centrality measures along with ISBC, while node labels are derived from diffusion models, particularly the Susceptible–Infected–Recovered (SIR) and Independent Cascade (IC) models. Various machine learning algorithms are applied to learn the relationship between structural features and spreading capability. Extensive simulations on multiple real-world networks demonstrate that the proposed framework improves prediction accuracy over traditional centrality methods. While machine learning models perform well when trained and tested on the same network, ISBC shows superior generalization in cross-network scenarios, consistently surpassing conventional methods.
Authors
- Murali Krishna Enduri (ORCID: https://orcid.org/0000-0002-9029-2187)
- Mondikathi Chiranjeevi (ORCID: https://orcid.org/0009-0008-1490-9406)
- V. Sateeshkrishna Dhuli
Institutions
- SRM University (IN)
Publication Details
- Journal
- Scientific Reports
- Published
- 2026-11-12
- DOI
- https://doi.org/10.1038/s41598-026-79588-7
- Primary Topic
- Complex Network Analysis Techniques
- Type
- article
- Field-Weighted Citation Impact
- 0.00