Which is the best document clustering open-source package?

Question

Which open-source package is the best for clustering a large corpus of documents? It should either decide the number of clusters by itself or it can also accept that as a parameter.

We have a large corpus of documents that don't really revolve around a particular topic - they are documents produced by sales and management people on various projects and clients in the organisation. I know that having such a spread corpus will degrade the performance, but we are trying to live with the best that we can get. Now, what is the best we can get :-)

Answer 1

A list of topic modeling software from the homepage of an expert in the field: http://www.cs.princeton.edu/~blei/topicmodeling.html

A competing leading group (with open source code): http://nlp.stanford.edu/software/tmt/tmt-0.3/

Another open source java project: http://mallet.cs.umass.edu/topics.php