Solorio’s research may help stop cyber-bullies, plagiarism or even terrorists

solorio sizedAs a world-renowned doctoral research university, UAB expects its teacher-scholars to win competitive grants and other awards to support their research. And they do not disappoint. This year, four professors have been chosen to receive National Science Foundation Career Awards — a prize the foundation describes among its most prestigious.

The NSF awards are provided to support the early career-development activities of professors who most effectively integrate research and education within the context of the mission of their organization. At UAB, those faculty include Thamar Solorio, Ph.D., and Ragib Hasan, Ph.D., assistant professors of computer and information sciences, Eugenia Kharlampieva, Ph.D., assistant professor of chemistry, and Karolina Mukhtar, Ph.D., assistant professor of biology.

The UAB Reporter will feature each of these outstanding researchers during the next few weeks, and provide insight into their work and its promise.

This week we focus on Solorio’s research and its potential national security reach.

Solorio received a $469,642 NSF Career Award for her research on authorship analysis in cross-domain settings. Her research focuses on improving the technology that will help extract characteristics from written documents that can help determine its authorship, generate a profile of the author or identify cases of plagiarism. Authorship analysis can be used for historical purposes, to settle disputes regarding the creators of a given document and to build a prosecution case against an online abuser.

Solorio is hoping to build models that can analyze writing from a variety of sources — including text messages, email, forum posts, blogs and social media updates — and find patterns to identify the author. In some cases, this is helpful to law-enforcement officials who need to identify the author of threatening online communications.

Q: What was your reaction to being selected for this honor by the NSF?

A: I was thrilled. It’s a very prestigious award so, it was a big deal.

Q: What are the current challenges in authorship analysis, and how does your research address them?

A: Most of the work that people do in authorship authentication is within the same genre. For example, people may only use emails to build computational models and test those models on emails as well. And that’s fine, because it has useful applications. It does, however, have limitations. When the data of interest comes from a different domain then that used to build the models, it’s very likely that the prediction performance you will get is too low to be useful.

Q: How did you become interested in this type of research?

A: I think it all started when I moved here to work at UAB. As a new assistant professor, you try to see what resources you have available. It’s a good idea to start exploring and exploiting the local resources. And we have a strong emphasis here in our department and at UAB on things related to information assurance, computer security and computer forensics. So, I tried to explore ways my area of expertise in natural language processing could align with these interests, and authorship analysis was a natural fit.

Q: How does your research tie in to national security measures?

A: This whole field of forensic linguistics has a lot of relevant applications in things like national security. For example, agencies dealing with national security may be interested in monitoring particular forums where terrorist organizations are known to be active. They may want to know who the leader is or who is writing particular posts. They may want to link a specific suspect to an email message that is discussing a threat or to a transcript of a telephone conversation. The hope is that you can use what we do in research and apply it to real cases.

Q: What does this award mean to you and for your research?

A: It’s a huge boost to my confidence as a researcher. It means my proposal was able to convince a panel of experts that this is a relevant problem to tackle, that the proposed solution is innovative, and that I have the potential to follow through and yield a relevant contribution to my field. It comes with a lot of prestige, and thus it also means a lot to me professionally. As a researcher, this gives me five years to really delve into this authorship-attribution problem. I’m just so excited I have these five years to develop this research. 

Research & Scholarship