This report presents a selection of my publications over the past ten years, conducted initially as an associate professor at ComplexNetworks team at LIP6 (Sorbonne University) before being a researcher at CNRS at ISP (ENS Paris-Saclay, University Paris-Saclay). My research mainly concerns the analysis and modeling of real-world networks, with a particular emphasis on multipartite structures.
The first chapter of my report illustrates this approach by showing how analysis and modeling of bipartite graphs are studied together in order to capture the real structure of complex networks and how this understanding can then be used to support other computer science tasks such as the ones in the field of information retrieval.
Two specific contexts of applications for the use of these approaches are then detailed. Firstly, the second chapter shows how the representation of networks as multipartite graphs, combined with constrained random walks and dispersion coefficients, can be used to quantify a notion of node diversity in networks. This theoretical approach is applied to several online music datasets in order to measure both the diversity of users and the effects of algorithmic recommendations from online platforms.
Finally, the third and last chapter reports on an interdisciplinary research conducted with legal colleagues. It aims at analysing corpus of legal documents using multipartite graphs in order to identify the citation patterns that lead judgments to become influential in the case law. Beyond the results in terms of network analyses of the different international courts, this part aims to show that qualitative and quantitative research are not antagonistic but, on the contrary, complement each other to provide a more comprehensive picture of the reality of court case law.
I conclude my report by discussing how the work developed in chapters 2 and 3, although very independent, converges towards a line of interdisciplinary research that seems to me very promising. It aims at showing how we can study the relevance, the feasibility and the place of computer science approaches in the legal field with the ambition of putting practitioners back at the center of expertise.