Analyse et modélisation de la structure multipartie des réseaux réels
In practise, real world depicts a large number of networks emerging naturally from many contexts (for example in computer science with router networks, satellite networks, web-page networks, in biology with neural networks, in ecology with biological interaction networks, in law with legal decision networks, in economy with interbank networks, in social sciences with social networks). All these networks, sometimes called real-world networks, have very specific peculiarities : they come from practical contexts, they are usually very big (sometimes with several billion of nodes and links). They also have statistical features in common : a weak global density coupled with a strong local density and we usually observe that nodes tend to be organized into groups, where they are more strongly connected between them than to the rest of the network, called communities. In this thesis, we are interested in the community structure of bipartite real-world networks. Firstly, we dedicated our work to analyze and characterize overlapping in the bipartite structure of four real-world networks by measuring local density of nodes. Then, we built up a community detection algorithm, called ComSim, for bipartite networks (and more largely for any unipartite and k-partites networks). Finally, our work also concerns community detection for multilayer networks by proposing an extension of Girvan-Newman modularity. In a nutshell, our method is capable of detecting both single-layer communities and multi-layer communities and can be easily plugged to any optimization algorithm.
Defence : 10/12/2018 - 11h Jury members : M. Éric Fleury, Professeur, ENS Lyon [Rapporteur]
M. Bertrand Jouve, Directeur de Recherche, CNRS, Université de Toulouse [Rapporteur]
M. Matthieu Latapy, Directeur de Recherche, CNRS, UPMC
Mme. Clémence Magnien, Directeur de Rercherche, CNRS, UPMC
M. Jean-Loup Guillaume, Professeur, Université de La Rochelle
M. Fabien Tarissan, Chargé de Recherche, CNRS, ENS Paris-Saclay