Mining for extremism: ASU tool helps predict potential threats
The 2013 Miss World competition was originally set to take place in the outskirts of Indonesia’s capital city, Jakarta. But two weeks before the event, hardline Islamic protestors took to the streets, denouncing it as a violation of their moral code and values. The Indonesian government, which had given permission to hold the event in Jakarta, changed its mind, forcing the organizers to find a new host site.
Islamic radicals succeeded in moving the competition to an island in southern Bali, called Nusa Dua, where the population is largely Hindu. Even so, the U.S., British and Australian governments issued security warnings to travelers in Bali, fearing that violence would erupt even in the new location.
Religious extremist groups, like some of the ones protesting the Miss World competition, are the focus of ongoing research at Arizona State University’s Center for the Study of Religion and Conflict. The project is funded through the Department of Defense’s Minerva Research Initiative, which seeks to understand the social, cultural, behavioral and political forces that shape various regions of the world. Since its inception five years ago, ASU’s Minerva project spans three continents and tracks the activities of hundreds of radical and counter-radical social movements.
“I think the real problem that we’re spotting is the intolerance these groups have against others that do not share their values or opinions,” says Hasan Davulcu, a computer scientist in the Ira A. Fulton Schools of Engineering. Davulcu leads the web mining team that collects online data related to the Minerva project, which looks at social movements in Southeast Asia, West Africa and Western Europe.
While Davulcu and his team are using technology to study foreign radical groups remotely, anthropologists are gathering data in the field. Mark Woodward, an associate professor of religious studies at ASU, spends much of his time in Indonesia, where he conducts ethnographic research in the local communities to find out what sorts of issues people care about. He gets a sense of when social movements have the potential to turn violent by getting to know the groups that oppose religious radicalism.
“These are indigenous local forces that are opposed to really nasty, violent people. These are the people that have to deal with the likes of Al Qaeda every day,” says Woodward, the lead researcher on the project.
Woodward and Davulcu communicate often, exchanging insights and data based on what they learn in their respective “labs.” For example, if Woodward hears from local university students that a radical group is planning a protest, Davulcu can analyze that group’s web presence and inform the anthropologists of any notable trends.
“It’s a synergistic relationship. If they find something that’s interesting online, we go check it out in person,” Woodward says. Combining local, in-person observations with notable trends online gives researchers a big-picture understanding of how social movements resonate in society.
Davulcu scours the websites of individual groups, their Facebook pages, their Twitter feeds and local news outlets, collecting keywords, hashtags and other bits of data. These are aggregated into a tool his team developed, called Looking Glass. The tool continuously gathers information, organizes it and displays it as an interactive graph, all in real-time.
Not only does Looking Glass track millions of tweets from nine different countries, it’s also multilingual, with the capacity to recognize any language. This allows the Minerva project team to study hundreds of social movements, ranging in size from small, 50-person groups to coalitions with more than 80 million followers.
How does the data mining team know what to look for online? They are informed by social scientists like Woodward, who has traveled to Indonesia, Malaysia and Singapore since the late 1970s, building rapport with locals and learning the issues that are important to them. To do that, he uses two basic research methods. The first involves conducting randomized surveys of target populations. The second is talking to people in the community. In the field of anthropology, this is called ethnography. Whereas political scientists tend to conduct formal interviews with organization leaders, anthropologists have informal conversations with a range of people in everyday settings, such as schools or mosques.
They also go to a lot of weddings.
“There’s always a sermon at the wedding, and it’s a chance to see a whole lot of people in one place,” Woodward says. “But the methodological trick to that is, you have to be invited to the wedding. You have to have good enough contacts and good enough relationships with the local community that you get invited.”
Through his ethnographic research, Woodward has found that communities in Indonesia are concerned about the same issues that people all over the world care about – the economy, the rising price of gas, religious freedom and maintaining a peaceful environment.
“Nobody likes to live in a place where there are bombs going off,” Woodward says. His research, combined with the Looking Glass tool created by Davulcu’s team, has provided valuable insights into the lifespan of social movements and how their influence changes over time.
While Looking Glass is very accurate and extensive in the data is collects, it cannot yet make logical connections between related events using its historical archive. For example, in northern Nigeria, many young people decry the corruption of local politicians. They see the solution to be a more rigorous adherence to Islamic law. In moments of political crisis and rapid social change, theological arguments take on greater importance and may give rise to both violent and non-violent religious movements.
“We would like the machine to be able to connect the dots and give us a summary, so this type of technology can become an assistant to a social science researcher,” Davulcu says. That’s his goal for the next stage of the project.
Still, the current capabilities of Looking Glass are no small feat. Davulcu’s team was invited to present four papers at the International Conference on Social Computing in September, with one of them winning “Best Paper Award” at the event. Two of the papers were also selected for publication in Human, a journal of the Academy of Science and Engineering. Davulcu and Woodward both presented at the Minerva conference held in Washington, D.C. in September.
“Not only do we have highly qualified expert ethnographers looking at these areas, but through the emergence of social media, now we can also see what grabs people’s attention online and follow what they are showing us to focus on the things that matter,” Davulcu says. “That’s the exciting contribution from the computer science perspective, to use a crowd-sourced filter to put the lens on interesting socio-cultural trends and developments.”
Mark Woodward is an associate professor in the School of Historical, Philosophical and Religious Studies in the College of Liberal Arts and Sciences. Hasan Davulcu is an associate professor in the School of Computing, Informatics and Decision Systems Engineering in the Ira A. Fulton Schools of Engineering. Both are faculty affiliates of the Center for the Study of Religion and Conflict, an interdisciplinary research unit of the College of Liberal Arts and Sciences that examines the role of religion as a driving force in human affairs.
Written by Allie Nicodemo, Office of Knowledge Enterprise Development