1
$\begingroup$

I'm trying to classify a set of 1656 tweets into different categories. I've read about different classification algorithms (supervised and unsupervised) but I'm really concerned because my set and the document text are small.

Which algorithm would you recommend for classifying small sets with small documents such as tweets?

P.S: I don't want to train a model with this data, I just need to classify this tweets into categories as an input for another task.

Thanks!! :)

Improve this question
$\endgroup$

1 Answer 1

Reset to default
0
$\begingroup$

Are 1656 tweets enough? That depends first and foremost on your task. See my answer here for more information.

Supervised or unsupervised?

Can you give (at least) one category for most documents? Then it is supervised.

Small sets vs large sets

For most scenarios in text classification, linear SVMs (Liblinear) work very well. Twitter data is somewhat harder to handle because of things like colloquial language, misspellings, abbreviations, hashtags...etc. However, if you clean and preprocess your data quite good, the algorithms work the same as usual.

Improve this answer
$\endgroup$

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.