Segmented Topic Model

From Cohen Courses
Revision as of 15:54, 29 September 2012 by Ymiao (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigationJump to search

Segmented Topic Model is a new form of topic model which can take into account the inner structures in documents.

Basic Ideas

  • As in LDA, one document d has a multinomial distribution v(d) over latent topics
  • In this document, each segment d,s (sentence or paragraph) also has a multinomial distribution over topics. This distribution is generated from a two-parameter Poisson-Dirichlet process r(d,s)~ Poisson-Dirichlet(v(d),a,b)
  • The topic label of each word is drew from the topic distribution of its segment

Citation

A Segmented Topic Model based on the Two-Parameter Poisson-Dirichlet Process. Lan Du, Wray Buntine, Huidong Jin. In Machine Learning, Volume 81 Issue 1, Pages 5 - 19, 2010.

Retrieved from "http://curtis.ml.cmu.edu/w/courses/index.php?title=Segmented_Topic_Model&oldid=12340"