Communication Dans Un Congrès Année : 2025

NetBone: A Python Package for Extracting and Comparing Backbones in Weighted Networks

Résumé

Backbone extraction simplifies complex networks by retaining their most significant structural elements. It plays a vital role in improving network analysis, visualization, and interpretation. However, selecting appropriate methods and evaluating their performance remain challenging due to the diversity of techniques and lack of standardized tools. We introduce Net-Bone, a Python package that integrates over 20 backbone extraction methods from statistical, structural, and hybrid categories. NetBone offers a unified framework for backbone extraction, filtering, evaluation, and visualization. It supports multiple filtering strategies and provides built-in topological measures. A key feature is its comparison framework, which enables users to benchmark methods using property-based metrics, progression analysis, distribution alignment, and consensus extraction. Through a practical example and visual comparisons, we demonstrate NetBone's flexibility, extensibility, and effectiveness in analyzing real-world networks.

Fichier principal
Vignette du fichier
FRCCS___Netbone___Ali__V2.pdf (1.97 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-05095444 , version 1 (03-06-2025)

Licence

Identifiants

  • HAL Id : hal-05095444 , version 1

Citer

Ali Yassin, Abbas Haidar, Hocine Cherifi, Hamida Seba, Olivier Togni. NetBone: A Python Package for Extracting and Comparing Backbones in Weighted Networks. French Regional Conference on Complex Systems, CSS France, May 2025, Bordeaux, France. ⟨hal-05095444⟩
273 Consultations
206 Téléchargements

Partager

  • More