I'd wager that there'll be at least some sort of Bayesian "hate" filtering. That means something like "Fuck McDonalds" won't trigger the ad.

No idea how they'd deal with positive-word, negative-connotation satire, though.

That suggests that there is an easy way to opt out: Just include some profanity each time you check in.

There's no way that they can implement thorough manual curation at the scale Facebook is at. Even better, as the parent suggested, use positive words to avoid any automated sentiment analysis: Starbucks makes my colon sing!

Sounds just like reddit ads.

I believe that business can bypass this to an extent by only opting to sponsor stories generated via applications...I read that in another article somewhere yesterday. runs off to find said article