We will be asked: "how does your notion of 'polarized text' differ from hate speech? How does it differ from posts that contain high 'anger' content? Aren't you really measuring the same sort of thing?" We need to be ready to answer such questions.
One possible way to answer this would be to compare the predictions of our classifier to the predictions of a hate speech or anger classifier, and show there are differences.
(Not sure who to assign this to yet; for now, I'll just put it on everybody's radar.)
We will be asked: "how does your notion of 'polarized text' differ from hate speech? How does it differ from posts that contain high 'anger' content? Aren't you really measuring the same sort of thing?" We need to be ready to answer such questions.
One possible way to answer this would be to compare the predictions of our classifier to the predictions of a hate speech or anger classifier, and show there are differences.
(Not sure who to assign this to yet; for now, I'll just put it on everybody's radar.)