A little experiment with Tensorflow. The script will fetch values from a toxicity classifier to judge if a string is 'toxic'. The original purpose of this neural net is to automatically flag toxic comments but here we are using it to give the toxicity a score, encouraging brief offensive phrases.
You can try it out here: https://driesdd.github.io/OffendedRobot/