Researchers develop tool that maps how hate speech spreads online

A team led by doctoral candidate Ehtesham Hashmi and professor Sule Yildirim Yayilgan at the Norwegian University of Science and Technology has built a tool designed to track and analyze online hate speech. The project included collaboration with the Western Norway Research Institute, police agencies and several Norwegian municipalities.

The tool creates a visual network of hate speech instances, linking statements to show where they originated, how they were picked up by others and how the conversations evolved over time. Hashmi said the goal is to help police and local governments direct resources toward prevention and awareness efforts where they can have the most impact, rather than simply prosecuting individuals.

The researchers trained the system on 14,000 sentences from Facebook, Reddit and Twitter, focusing on four key characteristics: ethnicity, gender identity, religion and sexual orientation. The tool classifies statements into three levels of toxicity and explains which words and phrases led to its decision, a feature intended to maintain transparency and trust.

Currently, the tool requires manual input, but the researchers are working to enable real-time automatic data retrieval from digital platforms. Yayilgan noted that the system is designed to improve with continued use, though human oversight remains essential to its ongoing development.