Common stuff. I work for Bazaarvoice; we have "generic" lists of "bad" words, as well as client-specific ones. We use them to automatically reject product reviews, or to at least flag them for human intervention.
You all can probably understand why these lists err on the side of false positive, rather than false negative.
There's also the occasional fun false positive, such as a reply to a review from a client that would include a URL with one of the "bad" words as a URL token: e.g., "To learn how to clean your vacuum, download the manual at www.manufacturer.com/download?product=123&ass=abc" - where "ass" is some abbreviated variable name that ends up causing replies to get rejected.
(There's also some really, really fun words in some of those lists that I have no idea how they got there. My favorite one is actually from a different company that maintains a word list: N.I.P.P.L.E.S. Yes, all capital, with periods in between the words. Highly specific, and I'd love to know what was the original trigger for adding that to the list.)
