2
$\begingroup$

I am not native English speaker and often times I use Ozdic to find the correct word choice. It's good overall but its database is quite limited. I think a tool based on Word2Vec should be able to do a great job for finding collocations/cooccurance.

With skip-grams, given a window size of n words around a word w, word2vec predicts contextual words c; i.e. in the notation of probability p(c|w). Conversely, CBOW predicts the current word, given the context in the window, p(w|c).

I am looking for a tool that takes a sentence with a blank spot as input and gives alternative options for the blank spot as an output.

$\endgroup$
  • $\begingroup$ Is there any reason why it has to include DL? $\endgroup$ – NBartley Sep 9 '17 at 0:10
  • $\begingroup$ Now that I think of it, not really. Perhaps I was under the influence of DL hype! $\endgroup$ – Thoran Sep 9 '17 at 20:48
1
$\begingroup$

You are describing the "Language Modeling" problem in NLP. Language modeling finds the probability distribution over sequences of words - given the words I have seen so far, what are the most likely words that will come next or are missing.

word2vec can find reasonable collocations/cooccurrence for individual words. However, word2vec will do a poor job with sentence completion because it does not model grammatical dependencies.

You can find a word2vec nearest individual word demo here.

| improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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