PhD ceremony Mr. O. Ersoy: Image-based graph visualization

PhD ceremony: Mr. O. Ersoy, 14.30 uur, Academiegebouw, Broerstraat 5, Groningen

Dissertation: Image-based graph visualization

Promotor(s): prof. A.C. Telea

Faculty: Mathematics and Natural Sciences

Software systems contain large and complex sets of dependencies between their components, such as call and inheritance graphs, and data flow and type dependency graphs. Analyzing such dependencies is arguably one of the most important tasks of software maintenance processes such as reverse engineering and reengineering.

However, understanding large sets of dependencies is challenging. Visualization is a method of choice, given the inherent difficulty for understanding large, abstract graphs. The visual depiction and understanding of large graphs, such as composed by the dependencies is an open challenge to both SoftVis and InfoVis communities. As the number of nodes and edges of a graph increases, node-link graph visualizations become challenged by clutter, i.e. unorganized groups of nodes and edges onto small screen areas. Clutter impairs tasks such as finding the nodes that a given edge (or edge set) connect, and at a higher level, understanding the coarse-scale graph structure.

Several approaches exist to reduce clutter in graph visualizations. One of the main directions for reducing clutter is edge bundling, or the geometrical grouping of edges that follow close paths. Edge bundling trades clutter for overdraw and produces images which are easier to understand and/or better emphasize the graph structure. Edge bundles can be rendered using various effects such as blending or shading. In this thesis, we take a new look at edge bundling layouts (EBLs) as a method for reducing the clutter in large graphs, with software understanding as an application area.