Good ideas and conversation. No ads, no tracking. Login or Take a Tour!
Yea, this one wasn't as interesting as the user one. The only thing I learned is that tag communities on Hubski aren't really well defined, perhaps due to the common ones that connect everything. In order to pick out groups of related tags, one could walk through the graph and delete the most common edges that are walked upon. That's basically the idea of betweenness centrality, and the general consensus is that if you remove the edges with the highest centrality, you're left with communities. It's an approach often applied to graphs when you want to analyze communities in a larger system.