Research Assistant for a Project at Worcester Polytechnic Institute

In this role, I worked in a team lead by Yichuan Li to research the linguistic tools and rhetoric used by disinformation, also known as "fake news." The research was applied to facilitate the development of a machine learning software that can distinguish legitimate and disingenuous news. The software was presented with articles my research partner and I found that had been extensively verified as fake or real, as well as linguistic tells we'd noted within those articles for lies (such as extensive use of "you," which often marks an attempt to establish affinity with the reader that is common to disinformation). Then, the software was prompted to try to classify each article itself. The machine learning software was built on a large language model (LLM) that included sentiment analysis as well as common vocabulary terms. The work I did for Li was accepted and featured at the 2021 IEEE International Conference on Big Data as an article I helped write and edit called "Multi-Source Domain Adaptation with Weak Supervision for Early Fake News Detection."

Later, I also helped Li research the concept of using machine learning software to help real news go viral— that is to say, to come up with ways to generate fact-based social media posts that are algorithm-friendly and appealing to various demographics of social media users.