English-only user filtering criteria
Hi, thank you very much for providing the English-only dataset!
If I understand correctly, this dataset is a subset of the 100M users dataset. I was wondering how you filtered for English-only users—what criteria or heuristics did you use to construct this subset? Also, does your filtering cover all 88M users in the original dataset, or only a subset of them?
Thank you again!
Hello, thanks for your question !
Actually it does come from the 100M users dataset, but i didn't process the whole, because that's way more than I need.
In fact, I just use this simple code :
def is_english_tweet(text):
lang, _ = langid.classify(text)
return lang == 'en'
And if I didn't make a mistake, I finally collected 250 tweets per user for the first 3000 users, so there are 750 000 tweets in total :)