|
I have a collection of documents, and know they may share a single topic. Is there a way to identify this topic? I know LDA (Latent Dirichlet Allocation) is an approach. But LDA result is to associate each document with a set of topics. However, I am more interested to find a common topic that can be used to label this collection. |
|
Do you mean "find a set of topics, and assign each document to one of these topics"? If so, this is just clustering, and you can do that with k-means or more complex approaches, as you wish. I would like to label this collection of documents with a single topic. I am not sure whether there exists an algorithm to achieve this goal.
(Mar 10 '12 at 09:39)
ouyang
|