Part of project: UIUC Affiliation Network

In my previous post post 3, I mentioned that I will only sample part of the 6,474 nodes and 8,271 edges of the network since d3.js was super slow at rendering the entire thing. While visualizing such large network is totally doable relatively quickly using Python or R, I wanted to explore some softwares specifically designed for the task.

I played around with two open-source software: (1) Gephi and (2) Cytoscape. There are pros and cons for both. I found the UI of Gephi to be very well-designed and intuitive as option is where you expect it to be. However, it seems very slow at rendering my network as the preview function with even forced direct layout took over 5 minutes (on a decent machine with 12mb of memory). I found Cytoscape to be the total opposite. It was very fast but the UI was not as intuitive. Embarassingly, it took me about an hour and a few YouTube videos to learn how to do simple things such as setting scaling node sizes and different node colors.

As it was originally an application for biological research, Cytoscape allows you to download some bio-databases to play around with which was very fun. The strong community support provided an ample number of layout types to experiment with. I eventually went with the strong-cluster layout from Allegro.

UIUC Universe
Click image to view full-size.

The orange dots correspond to people while the green dots correspond to the departments. Again, I made the size of the department nodes scale by it’s out-degree. Honestly, we cannot tell much by just looking at a screenshot so you probably should load the .cys file from the repository to Cytoscape to gather more information.