Comparative Analysis of Structural Backbone Extraction Techniques

Ali Yassin (1) , Hocine Cherifi (2) , Hamida Seba (3) , Olivier Togni (1)

1 LIB - Laboratoire d'Informatique de Bourgogne [Dijon]
2 ICB - Laboratoire Interdisciplinaire Carnot de Bourgogne
3 GOAL - Graphes, AlgOrithmes et AppLications

This study explores structural backbone extraction methods in diverse real-world networks, ranging from character to social networks. Eight techniques from the netbone pack- age are analyzed using Jaccard and Overlap coefficients to measure similarities. The Kolmogorov–Smirnov statistic evaluates weight and degree distribution preservation. Results highlight correlations and hierarchical relationships among techniques, with Doubly Stochastic outperforming in mimicking original network distributions. The study’s comprehensive insights contribute to refining and advancing structural filtering techniques, fostering a deeper understanding of complex systems.

Keywords: Complex Networks, Backbone Extraction, Filtering Techniques, Sparsification; Complex Networks; Backbone Extraction; Filtering Techniques; Sparsification

Domains: Computer Science [cs]

Ali Yassin, Hocine Cherifi, Hamida Seba, Olivier Togni. Comparative Analysis of Structural Backbone Extraction Techniques. French Regional Conference on Complex Systems, CSS France, May 2024, Montpellier, France. ⟨hal-04553418⟩