[Topic-models] Hyperparameters for DTM

Flavio Altinier Maximiano DaSilva fa269 at cornell.edu
Mon Jun 27 15:05:02 EDT 2016


Good morning,

I am trying to use the dtm code here (
https://code.google.com/archive/p/princeton-statistical-learning/downloads)
to study the evolution of topics on a dataset.

I am observing however that top-words on topics are almost not changing at
all through the years, and want to loosen the constraints on last-year
dependency. By checking the original paper (D. Blei and J. Lafferty.
Dynamic topic models. In Proceedings of the 23rd International Conference
on Machine Learning, 2006 [PDF
<https://www.cs.princeton.edu/~blei/papers/BleiLafferty2006a.pdf>]), I
believe it could be done by increasing sigma -- as seen in sections 2 and 3.

I tried checking the flags in the documentation but none of the sigmas
there are the flag I'm looking for. Which is the flag I should adjust? Or
is it hard-coded somewhere on the code?

Thank you,
Flavio.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cs.princeton.edu/pipermail/topic-models/attachments/20160627/ef6d9358/attachment.html>


More information about the Topic-models mailing list