The Profanity Power Index is a program that tracks the usage of profanity for specified targets on the Twitter public timeline, bundled with an interactive data visualization built with D3.js.
Naturally the target for this has been politicians. I've run it during several of the Republican presidential primary debates as well as during the general election debates between Donald Trump and Hillary Clinton. You can probably guess what those plots look like. In principle you can run this through any trackable term, but honestly no target's more fun than watching people curse at politicians in real time.
From a technical perspective there have been several iterations: the first was a Scala program that ran on Spark and wrote to HDFS, with the visualization built with a Clojure templating engine. Though there were a couple in between, the latest version is 100% Python, and writes to Elasticsearch.
Code (Python): https://github.com/timothyrenner/profanity-power-index
Code (Clojure): https://github.com/timothyrenner/ProfanityPowerIndex