If you look at table 4 it's pretty obvious all the model does is separate people who post about politics from those who don't. Given the methodology [0] this r

If you look at table 4 it's pretty obvious all the model does is separate people who post about politics from those who don't. Given the methodology [0] this result is unsurprising. Those labeled as "unreliable" seem to be universally political, while those deemed "reliable" are mainstream sources that may do stories on a variety of topics. [0] https://www.aclweb.org/anthology/P17-2102.pdf https://www.aclweb.org/anthology/P17-2102.pdf