This is cool, but a lot of PR is even more devious! At my first job ever, I was sometimes paid to simultaneously ghost-write stories by executives, then pitch the stories to magazines. So, it was still crap advertorials masquerading as content, but even harder to detect.

I wonder if you could do a linguistic analysis to gauge bias and ad-iness, and show a score. Gather a corpus of paid advertorials and compare to ostensibly unpaid material.