IU researcher hopes to predict, prevent crime with metadata

Post-doctoral researcher Emilio Ferrara has helped Italian police investigate murder, robbery, prostitution and drug trafficking.

Now, he’s at the IU School of Informatics and Computing and hopes to aid police in crime prevention by predicting individual cases of criminal activity through mapping out criminal social networks.

Everyone has their sphere of friends they communicate with on a regular basis, Ferrara said. By using social network theory, Ferrara and his colleagues have developed a program called LogAnalysis.

It can map out communication networks of a criminal organization using phone call data provided by police investigators.

“What we do is very different from what other government agencies — for example, NSA — do,” Ferrara said. “Our system is designed to analyze data of people who are suspects of some crimes or who are already known to the law enforcement agencies to be criminals.”

The system is not a broad, sweeping search through everyone’s phone data. Rather, LogAnaylsis only uses information obtained by police through court warrants, he said.
With a Ph.D. in computer science from the University of Messina, Italy, IU requested Ferrara join the School of Informatics and Computing in 2012. It was an honor, he said, considering the school’s national ranking.

U.S. News ranked IU as the 52nd best graduate school for computer science for the year 2014.

Ferrara said while analyzing phone activity during robbery investigations in Italy, he noticed spikes in phone calls before and after the robberies. By tracking the frequency of certain callers, middlemen in the criminal hierarchy could be identified.

Those who were higher ranking in the criminal hierarchy tended to make fewer phone calls, Ferrara said. Even callers who used a phone just once and threw it away were distinguishable.

At first, LogAnalysis might identify disposable phone users as multiple people. But after tracking a regular pattern, the program can determine that it is one person using multiple phones.

LogAnaylsis is focused on the analysis of mobile phone data right now, Ferrara said. In the near future, he said he hopes to incorporate data from online social platforms such as Facebook and Twitter as well as financial transaction activities.

“The goal is that we want to transform this platform in a predictive system, which means that we want to be able to not only study crimes as they unfold,” Ferrara said. “But also try to understand whether we can prevent or predict crime.”