Pages

Tuesday, January 10, 2012

Graphing wikipedia - update

I fixed up the crosslinking on our wiki grapher, so now each expanding node checks the list of existing nodes and connects them rather than creating a new node for every page it finds. A next step would be to use Wikipedias incoming links feature and also check node links in the other direction, however this would probably impact performance as a full web request needs to be made for every single node at creation, not just after it is clicked/expanded. Anyway, this fix already makes the resulting graph a lot more interesting:

By the way, this is centered around Mergesort and outgoing links are limited to 10 for testing purposes. I might keep it that way and then decide on a way to display the 10 (or so) most important links, although it'll be difficult to decide how to judge link importance.

No comments: