The graphs studied are heterogeneous multigraphs that may comprise thousands, hundreds of thousands or even millions of nodes, as well as a considerable number of edges. In fact, nodes are of different natures, such as abstractions, communities, parent elements containing child elements. Each node carries properties specific to its type, the concept it embodies and the information associated with it. Edges, on the other hand, describe existing relationships between nodes, with the possibility of several different relationships between the same pair of nodes. Edges also carry properties linked to the relationship they represent.
These graphs are therefore rich in information, but difficult for CAST Imaging users to exploit through graphical visualization. To ensure a good user experience, CAST needs to propose methods for navigating the application graph that guide the user in discovering and understanding the knowledge stored in it, without immediately overwhelming him with all the detail available.
The thesis will therefore focus on the study and development of advanced modeling and algorithms to produce simplified representations that can be used within current technical and economic constraints. These simplified representations will produce a clear and adapted visualization that will enable the user to perform a better visual analysis of the graph. This objective will be addressed by developing algorithms for improved performance in terms of analysis time (display, exploration, etc.), and comprehensible visualization (reduced and simplified representations). We will explore progressive visualization techniques so that the user can discover, on demand (interactively) or automatically, an application graph and its step-by-step representations, and furthermore display these parts of the graph from different angles with different information using, for example, 3D views of application graph representations.