Project Harmony is a new suite of tools in development that seeks to improve the tenor of online communications. How? Simply put, it increases our awareness of harmful language patterns. When visiting various social media websites (currently Facebook, YouTube, Google+, and twitter) content from posts and comments are scored by a comparison to our data set. When a certain post crosses a threshold of similarity to our data its color is changed making it more salient.
Project Harmony works because at the heart of the code is a ‘statistical model’ built from a growing codex of bullying statements. Harmony checks each block of text and gives it a score – too high of a score, and Harmony flags that bit of text. Project Harmony is pretty clever, but still not completely accurate. .. read more at