Sequential clustering with Carrot2?

Question

its possible to do "sequential clustering" with carrot2. What i mean with sequential clustering is, let say I have a set of documents that im clustering, and then after a some time new documents are ariving for clustering, is it possible to cluster new document with the old result or do I need to recluster all documents.

first post btw so be be gentle.

I don't know the answer, but what you refer to is usually called 'online clustering'. — micans, Mar 14 '14 at 16:54

score 0 · Accepted Answer · answered Mar 14 '14 at 21:43

Carrot2 does not support such operation mode out of the box. Depending on how frequently new documents arrive, you may be able to recluster a sliding window of documents and merge the newly created clusters with the old ones to get the complete result. I haven't tried such an arrangement before, so if it fails, you may indeed be better off with a dedicated online clustering algorithm.

Sequential clustering with Carrot2?

1 Answers1