Re:NeTV2 – An open video development board
by blackflame7000 on Saturday May 12, 2018 @10:12AM
(#17054143)
I'm thinking you could compile a list of commonly used words in positive or negative articles and count their frequency as a starting point. Also, you could search for nouns of importance like Trump or Obama and hone your bias detection.
On a more conceptual note, sentiment analysis detects (linguistic markers of) sentiment, not bias. Pretty much every news source will have positive sentiment for heartwarming human interest stories, and negative sentiment about atrocities. That's as you'd expect; we don't produce or consume news in a vacuum.