Closed aronwc closed 9 years ago
I have sampled the second thousand of tweets and labeled them. Also, I have reviewed the first thousand, that is why the values are different now. The values of labels got coincidentally very similar.
First sample 1K: positives = 353 negatives = 137 others = 510
Second sample 1K: positives = 323 negatives = 138 others = 539
Elaine has labeled a 1K sample, of which ~505 were positive towards e-cigs or indicated that the user used e-cigs.
She will sample 1K more and repeat.
If a user mentions seeing someone use an e-cig, we call this a "negative" example.