Dynamical detection of network communities

Autor(es)

Quiles, Marcos G. ^[1] ; Macau, Elbert E. N. ^[2] ; Rubido, Nicolas ^{[3, 4]}

Afiliação do(s) autor(es)

^[1] Univ Fed Sao Paulo Unifesp, Dept Sci & Technol DCT, BR-12247014 Sao Jose Dos Campos, SP - Brazil

^[2] Inst Nacl Pesquisas Espaciais, Lab Associado Computacao & Matemat Aplicada, BR-12227010 Sao Jose Dos Campos, SP - Brazil

^[3] Univ Republica, Fac Ciencias, Inst Fis, Igua 4225, Montevideo 11400 - Uruguay

^[4] Univ Aberdeen, Kings Coll, Inst Complex Syst & Math Biol, SUPA, Aberdeen AB24 3UE - Scotland

• Tipo de documento: Artigo Científico
• Fonte: SCIENTIFIC REPORTS; v. 6, MAY 9 2016.
• Citações Web of Science: 6

Resumo

structures. Specifically, communities are groups of nodes that are densely connected among each other but connect sparsely with others. However, detecting communities in networks is so far a major challenge, in particular, when networks evolve in time. Here, we propose a change in the community detection approach. It underlies in defining an intrinsic dynamic for the nodes of the network as interacting particles (based on diffusive equations of motion and on the topological properties of the network) that results in a fast convergence of the particle system into clustered patterns. The resulting patterns correspond to the communities of the network. Since our detection of communities is constructed from a dynamical process, it is able to analyse time-varying networks straightforwardly. Moreover, for static networks, our numerical experiments show that our approach achieves similar results as the methodologies currently recognized as the most efficient ones. Also, since our approach defines an N-body problem, it allows for efficient numerical implementations using parallel computations that increase its speed performance. (AU)

• Processo FAPESP: 11/50151-0 - Fenômenos dinâmicos em redes complexas: fundamentos e aplicações
• Beneficiário: Elbert Einstein Nehrer Macau
• Modalidade de apoio: Auxílio à Pesquisa - Temático
• Processo FAPESP: 11/18496-7 - Aprendizado semi-supervisionado dinâmico e ativo baseado em redes complexas
• Beneficiário: Marcos Gonçalves Quiles
• Modalidade de apoio: Auxílio à Pesquisa - Jovens Pesquisadores
• Processo FAPESP: 15/50122-0 - Fenômenos dinâmicos em redes complexas: fundamentos e aplicações
• Beneficiário: Elbert Einstein Nehrer Macau
• Modalidade de apoio: Auxílio à Pesquisa - Temático