In 2023, I worked with a team from the Paul Douglas Institute to investigate the linguistic markers of fake news. We used a dataset with 80,000 news articles, half fake and half real, and analyzed them using the Vader NLTK Lexicon for sentiment and the NRC Emotion Lexicon for emotion, as well as other analytical tools. We found that real news has more positive sentiment and tends to use longer words than fake news. This information can be used to help corporations and the public better identify fake news, hopefully curbing the spread of misinformation.

Below is our symposium presentation from 2023.