Skip to content
Skip to navigation menu


Can browsing Twitter help track and prevent crime?

21 August 2013


With hundreds of millions of online social interactions taking place each day, there’s a massive drive to harvest and use this ‘big data’, with implications for social and computer scientists, law enforcement agencies and governments around the world.

There are currently 400 million tweets each day on Twitter worldwide, of which COSMOS – the Collaborative Online Social Media Observatory – is able to mine, store and analyse between one and 10 per cent. The percentage may sound small, but this still equates to a colossal amount of data. By 2020, it is estimated that social scientists will have access to the largest data source on the planet, meaning the potential for research is enormous.

Dr Matthew Williams, Senior Lecturer in Criminology at the School of Social Sciences, says: “We have created a digital data observatory that automatically – and ethically – harvests data from social media and other open digital sources such as the Office for National Statistics. The data analysis tools developed so far for COSMOS include gender, location, frequency, network, topic, sentiment and tension analysis. These tools were initially validated in a pilot study that focused on racial tension on Twitter.”

Professor Omer Rana at the School of Computer Science and Informatics: “For academic use, and with permission from Twitter, you can mine up to 10% of this data each day for free, via the ‘gardenhose’ – the key challenge is how to store and scale the analysis of all this data. If you had the capacity to store and analyse it, even 100% of this data is available to use, but for the sum of around £40,000 a day!”

Having access to this type of information is a sea change to the kind of data social scientists are used to analysing. For example, the census is taken only every 10 years, and other surveys might be taken every one or two years. These new data are distinctive in capturing naturally occurring social interaction at the level of populations in near real time. Consequently they offer the possibility of studying social processes as they unfold.

Twitter in Numbers

Existing COSMOS projects are using this data to research the potential link between social media updates and crime, hate speech and suicide. So what about the direct uses for this kind of data, surely recent news regarding the National Security Agency in the USA and GCHQ in the UK is going to make people more suspicious of ‘organised snooping’?

Dr Pete Burnap of the School of Computer Science and Informatics: “A recent example of the use of this data was after the murder in Woolwich of Drummer Lee Rigby. We harvested 450,000 Woolwich-related tweets in the weeks following the incident. This enables us to analyse the opinion and reaction of those watching the news.

“Our research into racial tension around sporting events shows that there is a link between real-world actions and social tensions expressed via social media. Likewise, we have observed increased levels of cohesion between Twitter users in relation to the London 2012 Olympics. Through COSMOS we can now scientifically and empirically formulate and conduct social science experiments that are replicable by researchers across the globe.”

The COSMOS team are currently working with the Metropolitan Police to see if they can gauge and perhaps even predict crime in certain boroughs in London, and with Google to see if they can identify the spread of hate speech in social media.

“In the near future we plan to expand upon this work to look at potential links with voting behaviour, using the Scottish referendum as a case study,” Dr Williams continues. “Our ultimate aim is to establish an International Centre for Social Media Research that is fit for purpose for research in the 21st Century that will provide a much needed service to academics and policy and decision makers globally.”


The COSMOS team are Matthew Williams, William Housley, Adam Edwards, Jeff Morgan and Luke Sloan from the School of Social Sciences; and Pete Burnap, Omer Rana and Nick Avis from the School of Computer Science and Informatics.  

For more information visit and follow on the team Twitter @cosmos_project

This article was original featured in the Cardiff University Magazine for Alumni & Friends, Summer 2013. Available to view here: