Resumen
Social Network Analysis (SNA) provides critical support for policy formulation and complex decision-making processes addressing certain communities within larger populations, and community detection is a key prerequisite for this task. It allows identifying groups, mostly on static networks where the links between nodes are available. Beyond topological analysis, integrating node attributes and network dynamics yields more precise community identification. Two additional challenges remain insufficiently addressed in the literature: the detection of overlapping communities, since nodes frequently belong to multiple groups simultaneously, and the autonomous determination of the number of communities, which is typically treated as a fixed input rather than inferred from the data. In this paper, we develop an algorithm for Overlapping COmmunity DEtection in Dynamic Attributed NETworks (OCoDeDANet), which addresses the model selection problem through Automatic Relevance Determination (ARD). We test our approach on several dynamic networks where it outperforms state-of-the-art algorithms. Our experiments show the importance of considering the networks’ evolution and the nodes’ attributes. These elements are present in many real-world applications, such as workplace contacts and crime analytics as shown in our experiments.
| Idioma original | Inglés |
|---|---|
| Número de artículo | 134723 |
| Publicación | Expert Systems with Applications |
| Volumen | 334 |
| DOI | |
| Estado | Publicada - 1 feb 2027 |
Nota bibliográfica
Publisher Copyright:© 2026 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
Huella
Profundice en los temas de investigación de 'Detecting disjoint and overlapping communities in temporal node-attributed networks'. En conjunto forman una huella única.Citar esto
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver