A Binghamton University Ph.D. candidate has created a new framework that helps protect against and predict terrorist activity by identifying interconnected patterns in attacks.

Salih Tutun, a fourth-year Ph.D. candidate studying industrial and systems engineering, created the new framework as part of his dissertation. His framework considers terrorist attacks as interconnected events rather than individual occurrences, an approach that he said is often taken by counterterrorism agencies like the CIA, FBI and National Security Agency.

His paper, “New Framework that uses Patterns and Relations to Understand Terrorist Behaviors,” was published in the international open-access journal Expert Systems with Applications. It was co-authored by Tutun’s Ph.D. adviser Mohammad Khasawneh, chair of the systems science and industrial engineering department at BU, and Jun Zhuang, associate professor of industrial and systems engineering and director of undergraduate studies at the University at Buffalo. According to the paper, the framework is able to predict the facts of future attacks with 90 percent accuracy.

The framework operates by comparing aspects of multiple different terrorist events that occurred between 2003 and 2015. Tutun used data sets from 150,000 events, including hijackings, suicide bombings, armed assaults and assassinations. The tracked data consists of latitude, longitude, type of attack, weapon type and terrorist group responsible. By using these data sets, he was able to analyze the actions of terrorists and the direct responses to the attacks from various intelligence agencies.

In Tutun’s tests, the framework was able to accurately predict responses made by terrorist groups to intelligence agencies’ actions. By drawing connections between events and using the framework, Tutun created a theorized result of what had happened and then compared his theory with the actual event to test his predictions.

“With our framework, we understand what the patterns are and worry about how we can protect certain areas,” Tutun said. “And then, if the protection has failed, we are able to look at the event and determine whether or not it was a terrorist attack, and why it was successful.”

By analyzing past data sets, Tutun said he has gained an understanding of terrorist groups’ methods, as well as the ways they may be subconsciously learning. He also said that he thinks intelligence agencies can use the responses to unsuccessful attacks to better anticipate future attacks. In addition, Tutun said the framework could be even more efficient if connected with a search engine, which could gather larger and more complete data sets as events unfold.

“We are going to improve this framework and so it can work dynamically,” Tutun said. “If we pair this system with something like Google, it’s going to think dynamically. If something happens in the world it’s going to send an alert so that everybody who has it enabled can check and see what has happened that is different.”

While Tutun’s paper focuses exclusively on terrorist attacks, he said this framework can be applied to all kinds of social interactions, such as the decisions shoppers make in a store. By analyzing the purchases made, stores could predict what consumers will look for in the future. He claims that there are patterns in all facets of daily life and that the way people formulate interactions comes as a learned behavior, whether they are aware of it or not.

“It could be shoppers at Walmart, it could be diagnosis of disease, it could be patients at a hospital,” Tutun said. “If we collect the data for events, we can look at the patterns and analyze those changes to help figure out what’s coming next.”