MSc internship: interactive, visualized graph exploration for hypergraphs and traversals in Amsterdam

Context

In ING we have executed prior work where we explored the process of synthesizing middleware components based on a hyper graph of types and their relations. For more details we refer to this publication [0]. The work of this project will be executed in a research-oriented team [1].

Challenge

Although we have implemented a method that finds a path given the start-, and stop criteria as defined in the graph, visualizing the results is still very unsatisfactory. Even more importantly: if no results are found it is very hard and almost impossible to understand the ‘missing edges’ in the graph. Other directions, such as exploring hyperlink prediction [2], turned out to not be a viable solution to our problem.

We hypothesize that having a means of interactively exploring the graph, in conjunction with the search criteria, might lead to better results. As an example, you could think of a ‘Google Maps-like experience’ where you plan a route, optionally travelling via certain mandatory midway points etc, but this is by no means the definitive way of addressing this problem.

Expected project outcome

• Study current state-of-the-art means of visualizing (partial) hypergraphs

• Study and experiment with creating tooling to explore large hypergraphs

• Create a prototype exploration tool that aids in exploring large hypergraphs in

  conjunction with path finding constraints (e.g., starts at ‘x’, end at ‘y’, travel via ‘z’)
   

Candidate profile

We are looking for a student with the following profile:

• Like developing applications, especially on the JVM and in the browser

• Have a knack for (interactive) data visualization and exploration

• Are curious about graph theory

• Are a team player
   

[0]

[1] https://dev.azure.com/OmniSEA/_git/theses?path=/bin/sea.html&_a=preview
[2] https://dev.azure.com/OmniSEA/_git/theses?path=/2021/Manon_Wientjes.pdf