[Topic-models] Topicnumbers absent in output file

Ivan Savov ivan.savov at gmail.com
Wed Mar 16 11:50:23 EDT 2016


Hi Martha,



> I expected a file with a list with the columns docname, source, topic,
> percentage, topic, percentage, …, and so on.
>
However, in the file I get the topic numbers are absent.
>

> So I get a list with docname, source, percentage, percentage, percentage,
> …, … What do I do wrong?
>

Nothing wrong, you're doing the right things. This is what the output
format looks like:

  {{doc_id}} {{doc_path}} {{p(topic|doc) as a space-separated vector of
length T}}

An example topic model on 2 documents and 3 topic could look like this:

  0 file:document1.txt 0.3 0.2 0.5
  1 file:document2.txt 0.8 0.1 0.1

The topic index you're looking for is positionally encoded:  p(topic0|doc0)
= 0.3, p(topic1|doc0) = 0.2, etc...


Ivan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cs.princeton.edu/pipermail/topic-models/attachments/20160316/6cede8c7/attachment.html>


More information about the Topic-models mailing list