Have you looked at the original articles? Basically the question of how well model is doing is best answered by how well it models previously unseen data.
With lda you can directly compute the probability of the next symbol in a test sequence. I don't remember whether the mahout implementation does that, off hand. My guess is that there are better lda implementations around by now. Mahout was early with scalable lda, but surely there are other implementations by now as well. Sent from my iPhone > On Jul 29, 2014, at 21:17, vaibhav srivastava <[email protected]> wrote: > > Hi i am able run lda but wanted to know the accuracy for topic modeling or > any other efficiency measuement. > Thanks in adavnce > Vaibhav.
