Is there a variation of Latent Dirichlet Allocation (LDA) for grouped data? As an example, let us consider corpus of all Yahoo Q&A (where for simplicity we consider a question lumped together with all the corresponding answers as a document). While submitting a question to Yahoo's site, user has to specify a category to which it belongs to (let us, again for simplicity assume that categories are simple, instead of hierarchical). Is there a version of LDA that takes advantage of this information?
Asked
Active
Viewed 184 times
3
-
1LDA means linear discriminant analysis to many people here. Important to spell out what you mean. Locally famous terminology is not necessarily globally familiar. – Nick Cox Aug 08 '13 at 08:33
1 Answers
1
Have a look at Labeled LDA.
Ramage et al.: Labeled LDA: A supervised topic model for credit attribution in multi-labeled corpora
there is even an implementation available
benroth
- 305