Diploma and Master Theses (authored and supervised):

K. Unger:
"Interactive Visual Exploration ofLarge Bipartite Graphs usingFirework Plots";
Supervisor: E. Gröller, M. Waldner; Institute of Visual Computing & Human-Centered Technology, 2020; final examination: 2020-05-04.

English abstract:
In this thesis, we introduce a web-based interactive exploration interface for a broadaudience to investigate large, weighted, bipartite graphs. The motivation of this workis based on theMedia Transparency Databasewhich arises from an Austrian law thatcompels legal entities to announce their advertisement spendings to media organizationsand meets the specified characteristics.Most current interactive exploration tools use complex visualizations because they weredeveloped for domain experts. As the Media Transparency Database is of potentialinterest to a broad audience, we provide a framework not just for domain experts butalso for inexperienced users.Therefore, we conducted systematic benchmarks to compare state-of-the-art web-basedrendering techniques. Furthermore, we compared the performance of different librariesto determine the most efficient rendering solution and current limitations of web-basedrendering.We introduce the concept of Firework Plots, which aims to provide a common visualizationthat scales well with the size of the data. Our visualization concept is based on intuitivenode-link visualization in combination with multiple visualization and interaction concepts.Hierarchical aggregation is used to improve scalability. Constrained, layered, force-basedgraph layouts, as well as firework animations and seamless zoom, are used to allowinexperienced users to drill down the graph hierarchy and track nodes through thehierarchy. Moreover, visibility management is used to reduce clutter and improveperformance.Based on the insights of our web-based graph rendering analysis, we implementedour framework and the concept of Firework Plots. We show the usefulness of theimplementation by discussing different use cases and comparing it to related work.Moreover, we conducted multiple benchmarks to show the rendering performance and calculation times.

Electronic version of the publication:

Created from the Publication Database of the Vienna University of Technology.