7

I'm looking for a dataset for moods or emotions (Happy, Angry, Sad) classification. That's to classify the sentiment of a given text. I would like to use Naive Bayes classifier for this analysis. Not only to train and test the model with the dataset, but rather to practice doing sentiment classification. Do you suggest any resources?

4

Some nice data sets for practicing sentiment classification are:

These are some open datasets which contain emotions like happy, sad, etc:

  • 3
    Some useful links, but be aware that the first one says it's not open. (and it's code, not a dataset). The second one's a broken link. Many of the others don't explicitly list a license. (eg, the third said 'free' but that doesn't mean it's 'open'). – Joe Nov 16 '15 at 10:22
  • @Joe Thanks for pointing out. I have used these in my projects. Some of them can be obtained by e-mailing the compilers. But yeah, I think some aren't open here. – Dawny33 Nov 16 '15 at 10:24
  • Actually I am looking for some text data which will contain Happy, Angry and sad related corpus not positive/ negative and not some word data – SOURAV Nov 18 '15 at 5:39
  • Thank you for letting me know. Added another relevant data source. The data is in the form of .sav files. – Dawny33 Nov 18 '15 at 5:47
  • 1
    Good answer (+1). Could you also please edit into your answer which ones are open and which ones not? – eigenvector Dec 13 '17 at 5:23
2

This would be a good data source and the researcher also done a work on it. refer it too.

good luck

  • Actually I am looking for some text data which will contain Happy, Angry and sad related corpus not positive/ negative and not some word data – SOURAV Nov 18 '15 at 5:39
1

Why not you use Twitter Search API to search your particular type text, then do some text modification. Sorry, I am not expert, it is just an idea.

1

The best one I know is this:

https://www.cs.cmu.edu/~./enron/

this is also available in MySQL format!

https://www.cs.purdue.edu/homes/jpfeiff/enron.html

It's the email corpus of Enron during the moments when it collapsed. The emails are clearly characterized by underlying feelings.

This dataset is quite unique and heavily described in the podcast Linear Digressions. It's one of a kind because after privacy laws kicked in it is not possible to ever get anything else like it again.

Happy Sciencing!

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.