Social media brings out the best and worst in people, but more often the worst. For social scientists, however, it’s a golden age of studying aggressive human behavior and the psychology behind online spats. A team of Finnish researchers is currently analyzing different characteristics of fights on Twitter — how they originate, how they spread, and how to better visualize them for data purposes. And results actually look pretty cool — making those emotional outbursts look like colorful firecracker explosions.

How the group visualized fights on Twitter. See the text for an explanation of what it means.

In an unpublished paper uploaded to the academic repository arXiv, Kiran Garimella and her colleagues at Aalto University in Helsinki illustrate the traits of controversies on Twitter, and how they compare to noncontraversial conversations.

Garimella compared the keywords, hashtags, and networks of retweets/favorites that followed heated tweets, versus benign or friendly tweets. They studied these clustered networks and observed how the polarized sides form and communicate with one-another, and mapped out a few examples.

In the above picture, the A and E graphs illustrate the hashtag #beefban (related to the contentious proposal to ban beef in India), and B and F represent #russia_march (used in discussions of Russia’s invasion of Ukraine). On the other side, C and G illustrate the much-less controversial #swsw, and D and H demonstrate the network for #germanwings. The polarity and intensity of debates are clear in the first four, while the latter four show a much more homogeneous network.

The Finnish researchers are still working on developing better tools to measure polarization in Twitter controversy, but this is a great first step towards the creation of models that could help scientists or those in media identify social media fights before they blow up, and be better prepared to follow and report on them.