LDA with topicmodels package for R, how do I get the topic probability for each term?

Question

I'm using the topicmodels package for LDA. I would like to create a visualization that shows how related or non-related each topic is. I envision a cluster of words that are unique to topic 1, but with a few keywords that are shared connecting to another topic. Any advice here would be great. To continue:

To do this, I need to know the each term probability to each topic. How do I get this with the topicmodels package? I can view the terms with:

terms(LDAmodel, 15)

But I don't know how to get values. Ideas?

score 1 · Accepted Answer · answered Jan 25 '16 at 22:32

You can use posterior()$terms to get the posterior probability for each term. posterior()$topics gives the probability for documents.

Example adapted from help(LDA):

data("AssociatedPress", package = "topicmodels")
lda <- LDA(AssociatedPress[1:20,], k = 2)
terms <- posterior(lda)$terms

## posterior probability for the first 5 terms (alphabetically)
terms[,1:5]
         aaron      abandon    abandoned   abandoning       abbott
1 3.720076e-44 3.720076e-44 3.720076e-44 3.720076e-44 3.720076e-44
2 3.720076e-44 3.720076e-44 3.720076e-44 3.720076e-44 3.720076e-44

LDA with topicmodels package for R, how do I get the topic probability for each term?

1 Answers1