Rohan Phadte: Fake news fighter
Rohan Phadte is a third-year electrical engineering and computer science student on a mission to battle fake news with the power of machine learning. Phadte, along with his roommate, collaborator and computer science major Ash Bhat, created Robhat Labs as a side project to develop software that tracks how false or misleading information spreads across the internet.
“It’s hard to determine if something is fake news,” Phadte said. “Machine learning algorithms can’t tell if something is true or not.” But what is possible, he said, is tracking patterns of how things are shared. And by using waypoints and benchmarks, machine learning technologies are able to tell if information is coming from legitimate sources, or sources devised to spread misinformation.
The Robhat Labs team first grabbed their own headlines in 2017 when they built NewsBotAI, a Facebook Messenger app that can identify political slant in news articles. Later that year, the team built a Google Chrome extension, botcheck.me, which adds a clickable button to Twitter profiles, enabling users to vet whether the handle is run by an actual person. Unlike other social media profiles, Twitter allows users to create bots, or automated functions, which are used for all kinds of things, ranging from users optimizing when their posts are published, to more nefarious activities like creating armies of fake profiles to quickly spread propaganda.
Phadte said the team’s work has been well received. They’re now in the process of developing more advanced machine learning algorithms that will help legitimate content producers secure their photo and video content. One such algorithm can analyze the features in visual assets and create a dense representation, or a compressed version of the binary information unique to the image or video. Producers can then get alerts if their content is being hijacked or manipulated online.
The team is working with political groups and media outlets as the battle against fake news rages on.
< Back to previous page