You can click through the full Nature article, which is nice. I haven't read the article in depth, but it looks like interesting stuff. A bunch of models in Bay

You can click through the full Nature article, which is nice. I haven't read the article in depth, but it looks like interesting stuff. A bunch of models in Bayesian non-parametrics are, like the described model, based on Polya urns. It should be a simple matter to use this model for some interesting clustering tasks, perhaps as an alternative to the CRP.