A SAT-based framework for overlapping community detection in networks - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2017

A SAT-based framework for overlapping community detection in networks

Résumé

In this paper, we propose a new approach to detect overlapping communities in large complex networks. We first introduce a parametrized notion of a community, called k-linked community, allowing us to characterize node/edge centered k-linked community with bounded diameter. Such community admits a node or an edge with a distance at most k2 from any other node of that community. Next, we show how the problem of detecting node/edge centered k-linked overlapping communities can be expressed as a Partial Max-SAT optimization problem. Then, we propose a post-processing strategy to limit the overlaps between communities. An extensive experimental evaluation on real-world networks shows that our approach outperforms several popular algorithms in detecting relevant communities
Fichier non déposé

Dates et versions

hal-01696118 , version 1 (30-01-2018)

Identifiants

Citer

Saïd Jabbour, Nizar Mhadhbi, Badran Raddaoui, Lakhdar Saïs. A SAT-based framework for overlapping community detection in networks. PAKDD 2017 : Pacific-Asia Conference on Knowledge Discovery and Data Mining, May 2017, Jeju, South Korea. pp.786 - 798, ⟨10.1007/978-3-319-57529-2_61⟩. ⟨hal-01696118⟩
219 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More