If you’ve ever gotten halfway through a “news” article only to realize it was an opinion piece, you’re not alone. With the advent of a commercialized 24-hour news cycle, news organizations have had to come up with ways to keep the ad revenue flowing. Opinion-based segments on news channels offered organizations a way to make “news” happen on a schedule and sell ads accordingly. Catering news to specific viewpoints has become a billion-dollar business, and the line has continued to blur between standard reporting and opinion. But what if you just want the facts without all the propaganda and opinions? Two students at the City University of New York (CUNY) are using their ACCESS allocation to train artificial intelligence on the difference between opinion and fact when it comes to the news.

Using Purdue University’s supercomputer Anvil, Vivek Sharma, a student at the CUNY Graduate Center and a lecturer at the John Jay College of Criminal Justice, and Mohammad Shokri (CUNY) worked on two separate research projects to determine how the news was being presented. One project focused strictly on determining if a news source used propaganda, while the other project focused on training machine learning models to detect opinion or subjectivity, as opposed to objectivity or facts-only news.

Sharma and Shokri used Anvil’s computational power to speed up their work. “Training these models requires GPUs and a lot of memory,” says Sharma. “Anvil gives us both of these things, which is great. Trying to train these models on your laptop is unimaginable.”

When you get an allocation through ACCESS, you’re getting more than just hardware – the resources provided through ACCESS often come with applications and software packages key to analyzing your data, something Sharma was quick to praise when speaking of his work.

Training these models requires GPUs and a lot of memory. Anvil gives us both of these things, which is great. Trying to train these models on your laptop is unimaginable.

–Vivek Sharma, CUNY Graduate Center student

“Another good thing about Anvil is that there are so many packages already on the system,” Sharma said, “so it’s super ready for any kind of machine-learning task. Most of the time, any package I needed to run an ML task was already on Anvil. Beyond that, Anvil’s support team was great. Anytime I needed help, I emailed the support team and would receive a response really quickly. I even told my supervisor about this, and she was impressed, as other systems that she’s worked on would take days to respond.”

If you have research that could benefit from HPC resources, go here to get started with ACCESS. You can read more about this research here: Researchers use Anvil supercomputer to help detect manipulation in media

Resource Provider Institution(s): Purdue’s Rosen Center for Advanced Computing (RCAC)
Affiliations: City University of New York
Funding Agency: NSF
Grant or Allocation Number(s): CIS230306

The science story featured here was enabled by the U.S. National Science Foundation’s ACCESS program, which is supported by National Science Foundation grants #2138259, #2138286, #2138307, #2137603, and #2138296.