Visualisation of Overlapping Sets and Clusters with Euler Diagrams. (Diagrammes d'Euler pour la visualisation de communautés et d'ensembles chevauchants)

In this thesis, we propose a method for the visualisation of overlapping sets and of fuzzy graph clusterings based on Euler diagrams. Euler diagrams are probably the most intuitive and most used method to depict sets in which elements can be shared. Such a powerful visualisation metaphor could be an invaluable visualisation tool, but the automatic generation of Euler diagrams still presents many challenging problems. First, not all instances can be drawn using standard Euler diagrams. Second, most existing algorithms focus on diagrams of modest dimensions while real-world applications typically features much larger data. Third, the generation process must be reliable and reasonably fast. In this thesis, we describe an extended version of Euler diagrams that can be produced for every input instance. We then propose an automatic procedure for the generation of such diagrams that specifically target large input instances. Finally, we present a software implementation of this method and we describe some output examples generated on real-world data.