We have a requirement to have comments on an application auto-moderated for obscenities etc. There will still be a real world moderating looking at comments, but they want to filter out the really bad stuff automatically.
What would be the best way to do this?