2
$\begingroup$

I want to use a neural network to do some topic analysis in a textual corpus. I have used neural networks before where there is a clear decision boundary between the category to which some observation belongs. However, in topic modeling, a single document can have multiple topics present simultaneously.

So I want to know if I can train a neural network model to associate multiple topics to each document in the corpus. I suppose this could be some sort of vector that just indicates the probability of some topic existing in that document>

I was looking around for any papers on this topic, so if there are any citations that you want to pass along, that would help too.

$\endgroup$

1 Answer 1

2
$\begingroup$

Yes you can use neural networks in this way - it's an example of multi-label learning. There are many approaches, but one is to create binary target vectors indicating which labels should be applied to each document. Or a vector of probabilities as you suggest.

There are plenty of papers on multilabel learning, with or without neural nets - one recent review is Zhang & Zhou 2014.

$\endgroup$
1
  • $\begingroup$ Wow, this article was a great resource. Thanks so much for the insight. $\endgroup$
    – krishnab
    Jul 28, 2015 at 19:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.