ModeRad helps identify potentially radical and extremist content on Twitter.
It is a desktop application developed as part of the PERICLES project.
The tool was built using machine learning. The machine learning model was trained to identify content demonstrating indicators of radicalisation based on a taxonomy developed by CRIMINA members.
The tool is based on analysis of metadata about violent and radicalised speech on the Internet. It analyses the incidence and prevalence of these phenomena, its typologies, and its metadata factors.
The software detects and records the violent and radicalised discourse on Twitter based on various metrics, including:
- The content of Tweets
- Images included in the Tweets
- Number of followers of the Twitter user
- The URLs linked to in the Tweets
- How much the Tweets have been shared/retweeted
After running the tool, law enforcement offers are given a filtered list of ‘high risk’ tweets that they can review.
The tool only summarises Tweets that indicate direct or indirect incitement to physical or moral violence.
The aim of the tool is to:
- Detect radical content to remove it
- Identify radical content users
- Monitor suspicious hashtags
This video provides a further introduction of the tool.