Pages

Tuesday, January 10, 2012

Visualizing Pagelinks

In this blog's last post I talked about force graph balancing and some of the things it can be used for, more specifically visualizing networks such as links between webpages. Robert BrĂ¼nig and I are currently working on a small projects that does just that, by scraping pages (currently Wikipedia articles) for links and then representing them graphically. It's written in C++ using the Qt framework. This would've been fairly simple in Python, however for larger networks the program becomes very computationally intensive, and we wanted to get good performance. Besides, using C++/Qt allows for much more stable deployment, and Qt has a lot to offer in terms of modules (widgets, graphics, web access and such). Screenshot below.

As you can see, the graph quickly grows quite large for wikipedia pages due to the high number of pagelinks on each article. We're thinking of focusing on Youtube instead, visualzing a gives clips related videos - perhaps we might even be able to embed the videos in the nodes, so that they can be watched directly in the program.

There are also still some issues with the graph balancing algorithm, we still need to find a good combination of attraction/repulsion/weighting parameters that produces a smooth and stable graph that balances quickly.

No comments: