So my problem is as follows: I get responses (such as "yeah whatever", "yes do it", "no don't do it", "nah", "yeah do it" etc.) and I need to classify them into either "yes" or "no" i.e. a binary classification. What algorithm and features should I use for this?

My thoughts yet: Use an averaged perceptron or logistic regression or boosted trees. As for features, I was considering using the tf-idf score and sentiment analysis but I can't think of much else.

  • 2
    $\begingroup$ I mentioned several other candidate features in my answer to your last question. Did you spend some time learning about those? I'd recommend spending more time following up on the leads you already have before asking for more. There are lots of techniques in the NLP literature, and there's little point in us repeating standard material, so you'd probably be better off spending some quality time learning about NLP methods. $\endgroup$ – D.W. Jun 4 '16 at 10:14
  • $\begingroup$ @D.W. Thanks again for your response! You mentioned two of them - bag of words and word2vec and said "That belongs in a separate question" - I'm sorry if I misunderstood you! While I have studied bag of words and word2vec in the past, I was wondering what additional features I should use here specific to yes/no classification. Please could you and the others help me out with that? [I edited the question too for another idea that I found after speaking t someone I know!] $\endgroup$ – Mathguy Jun 4 '16 at 20:26
  • $\begingroup$ Your thoughts consist of a list of things. Why do you think would these be suited for your task? What distinguishes them, and which would you prefer and why? $\endgroup$ – Raphael Jun 4 '16 at 22:12
  • $\begingroup$ @Raphael thank you for your response! I would likely try a variety of methods and see which one performs the best, but I'm leaning towards boosted trees as they're relatively simpler but still work very well in practice. [homes.cs.washington.edu/~tqchen/pdf/BoostedTree.pdf] $\endgroup$ – Mathguy Jun 4 '16 at 23:45
  • $\begingroup$ its a speech classification type problem. there are many std approaches. it would be better if you give more detail. drop by Computer Science Chat for more response. also Latent Semantic Analysis / Latent Semantic Indexing are often used for similar purposes. $\endgroup$ – vzn Jun 5 '16 at 2:46

Your Answer

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

Browse other questions tagged or ask your own question.