Law enforcement may target hate crime by analyzing Twitter

A new tool will try to connect hate speech with in-person crime.

Cardiff University is currently working on a project to let law enforcement scan social media to identify outbreaks of hate crime. According to the Financial Times (subscription required), this $800,000 project is being funded by the US Department of Justice; its main component is an algorithm that scans Twitter to identify hate speech in defined geographic regions of the US. From there, the hope is that the algorithm can look for patterns between hate speech online and violent actions offline.

Eventually, police may be able to use the tool to predict violent outbursts and hate crimes taking place after "triggers" like this week's police shooting in Charlotte that left Keith Lamont Scott dead. What's perhaps most notable about this project is that the algorithm will use machine learning to pick up less obvious words associated with hate speech. "It doesn't always have to use derogatory words associated with racism: it could be much more nuanced, which is the major challenge in the project," computer scientist Peter Burnap from Cardiff University said to the Financial Times. "We are using natural language processing to identify cyber hate in all its forms."

Los Angeles will be the first cit to test this new tool, and the goal is to help police peacefully intervene before any violence occurs. It's still an ongoing process to teach the algorithm how to successfully identify trigger phrases -- over the next three years, it'll analyze tweets in the context of specific events (like the upcoming US presidential election), map them to areas of different cities and then see how those tweets map up with reported hate crime in that area. With Twitter still not doing much to curb the rampant abuse that takes place on its platform, it seems like the Cardiff University researchers will have plenty of data to analyze.