Large graphs are challenging to visualize and interactively explore, especially on the web. In a master thesis, a prototype of Firework Plots for large interactive graph on the web has been presented . Using Firework Plots, groups of nodes are dynamically aggregated and can be interactively expanded by zooming into the graph. Rendering is performed using three.js.
The Firework Plots demo shows the content of the Media Transparency Database . This database contains all advertisement expenses that have to be reported by public authorities in Austria since 2012. This dataset is of high public interest, as it reveals in which media public companies, but also political parties, spend tax money for advertisement. It is therefore regularly parsed by journalists . However, it is hard for the general public to access and explore the entire data set.
Currently, Firework Plots can provide some overview, but the graph visualization is still cluttered. In this project, students should integrate state-of-the-art edge GPU-based edge bundling techniques (such as CUDA-  or texture-based edge bundling ) to make the visualization faster and more aesthetic. Optionally, other optimizations (e.g., concerning the layout or the labeling of the nodes) can be performed. The goal is to increase the user experience so that the data can be made accessible to a wide public audience. This work can also be done in a group of two.