Mining a medieval social network by kernel SOM and related methods

Nathalie Villa (1) , Fabrice Rossi (2) , Quoc-Dinh Truong (3)

1 IMT - Institut de Mathématiques de Toulouse UMR5219
2 AxIS - Usage-centered design, analysis and improvement of information systems
3 IRIT-MELODI - MEthodes et ingénierie des Langues, des Ontologies et du DIscours

This paper briefly presents several ways to understand the organization of a large social network (several hundreds of persons). We compare approaches coming from data mining for clustering the vertices of a graph (spectral clustering, self-organizing algorithms. . . ) and provide methods for representing the graph from these analysis. All these methods are illustrated on a medieval social network and the way they can help to understand its organization is underlined.

Mots clés: social network; large graphs; SOM algorithm; graph drawing; clustering; spectral clustering; heat kernel

Domaines: Applications [stat.AP]; Statistiques [math.ST]; Histoire; Méthodes et statistiques

Nathalie Villa, Fabrice Rossi, Quoc-Dinh Truong. Mining a medieval social network by kernel SOM and related methods. MASHS 2008 (Modèles et Apprentissages en Sciences Humaines et Sociales), Jun 2008, Créteil, France. ⟨hal-00278196⟩