Khayrallah Center Thinks Bigger With Research Facilitation Service

Since the Khayrallah Center first released its Arab American Newspaper Database in 2020, the collection has grown to nearly 500,000 pages of historical texts.

The database helps fill a digital gap in Arab American archival material, which is challenging to digitize with common optical character recognition software. 

“We developed our own machine learning system that can recognize this handwritten text,” said Mehreen Saeed, a Khayrallah Center research associate. “This is our ultimate goal — to have a system where we can converse in any language we like and be able to extract information from different types of documents.”

The Khayrallah Center’s machine learning model has opened the door for not only more intelligent search capabilities, but also more historical data.

“We’re ingesting a lot more data and we’re also getting a lot more traffic. It’s been an exponential growth and, equally, the cost has been exponential growth,” said Akram Khater, the center’s director. “We want to continue to be open source. We want to continue to actually innovate and create new ways in which people can use the data.”

Khater shared his concerns about rising cloud computing costs during a consultation with the Research Facilitation Service (RFS). The RFS is a collaboration between the Office of Information Technology, the NC State University Libraries and the Office of Research and Innovation that serves as a single point of contact for researchers to learn about available resources.

In addition to helping the Khayrallah Center find a solution, the RFS team recognized an opportunity to better understand additional resources that could benefit many researchers at NC State.

“We’d been looking into nationally funded, broadly available resources that are hosted at other universities and made available to researchers around the U.S., namely through this program, ACCESS,” said Derek DeVries, an RFS research integration consultant.

Funded by the National Science Foundation, ACCESS provides researchers and educators with advanced computing resources at no cost. Options like these are welcome as NC State’s need for computing capacity over the next five years is estimated at 10 times the current amount.

Together, the Khayrallah Center and the RFS explored possible solutions before choosing the best one, ultimately migrating some of its servers and data storage from Amazon Web Services to the Indiana University-hosted Jetstream2 platform.

“They needed the GPUs to power their workload and the storage to hold their data,” said DeVries. “We were able to help them with the process of getting an allocation. They tried it out and it seems to have worked really well for them.”

Mahim Dashora, a College of Engineering graduate student and developer for the Khayrallah Center, handled most of the migration. He said it felt good to see the cost savings while maintaining reliable services.

“I had a nice learning experience with it. As a student, I am getting opportunities to learn more about cloud and systems technology and how to solve more problems that are having real-world impact,” said Dashora.

Khater said the RFS’s hands-on, collaborative approach was comforting when working through the technical details.

“Sometimes this stuff is really kind of overwhelming, to be honest. The options are overwhelming, the data security is overwhelming. And so I think what was really helpful was for them to act not only as a connecting point between that group and our group, but also as translators in some ways,” said Khater.

While the RFS assists researchers across the university, it has officially rolled out its services to the College of Humanities and Social Sciences, College of Sciences, College of Agriculture and Life Sciences and College of Natural Resources.

For Khater, collaborating with the RFS means his team at the Khayrallah Center can think bigger.

“This was not a one-time consultation, but it’s a relationship,” he said. “And I really think that’s what it means for us — it’s knowing that, within the university, we have an office that is dedicated to helping folks who are working with data be successful.”

This post was originally published in Office of Information Technology News.