2
$\begingroup$

I want to give labels to different topics created using LDA. I don't want to do it manually. I saw some papers on automatic labeling but I am still confused.

How can I use the information produced by LDA to automatically generate names to the topics generated?

Improve this question
$\endgroup$
0

1 Answer 1

Reset to default
0
$\begingroup$

I suppose it is late to answer, but it may turn out to be helpful for others.

LDA presents all topics as a mixture of keywords with their weights For example, in gensim implementation it looks like this:

topic #0: 0.009*river + 0.008*lake + 0.006*island + 0.005*mountain + 0.004*area + 0.004*park + 0.004*antarctic + 0.004*south + 0.004*mountains + 0.004*dam

So you may name a topic using the keywords with the largest weights (like River+Lake). Or you may use something like WordNet to find the most common hypernym for them. For example, in python you can do it like this.

from nltk.corpus import wordnet as wn
wn.synset('river.n.01').lowest_common_hypernyms(wn.synset('lake.n.01'))
Improve this answer
$\endgroup$
2
  • $\begingroup$ i am not working with English so the option of wordnet is not suitable in my case. $\endgroup$
    – user3778289
    Commented Aug 28, 2018 at 18:05
  • $\begingroup$ There are wordnets for other languages, search for Multilingual wordnet with your favorite search engine. $\endgroup$
    – user10169
    Commented Dec 21, 2018 at 18:26

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