COVID-19 is a large-impact historical event that sent many of us online in the absence of in-person events, specifically online to social media platforms. Miranda Phillips is an FSU student and UROP (Undergraduate Research Opportunities program) participant who dissects the intersection between social media and COVID-19. The primary focus of her research is using machine learning to track emotional responses to COVID-19 using social media platforms like Twitter.
This ongoing project began by analyzing tweets and creating a deep neural network to determine which emotion is being portrayed through any selected COVID related tweet. Phillips assisted in creating model software that extracts which emotion a tweet portrays and sorts it accordingly. This model categorizes how people in the United States responded to the pandemic and provides data that would be helpful to public officials when deciding how to handle similar situations.
RCC resources such as Jupyter Notebook and Open OnDemand aided in creating this model. Efficiency is key when it comes to processing high volumes of data, such as all tweets relating to COVID-19 in the height of a global pandemic. Using RCC resources, about 30 million tweets have been analyzed within this project. Ideally, this model will progress to analyzing any large cluster of social media topics.
“Due to high computing performance, we were able to get the computer to do a lot that we couldn’t do with our own resources because of the extensive capabilities of the RCC,” states Miranda Phillips.