A central feature of the book is our analysis of the degree of positivity or negativity of a newspaper articles. To perform this analysis, we developed a new sentiment analysis method, which outperforms other lexicon-based approaches on standard benchmarks and is more robust across a range of different domains.
STAIR Lab is a computational social science lab conducting collaborative research with undergraduate students at William & Mary. We use natural language processing (NLP) tools to study how politicians and the media (both traditional and social media) think and talk about international issues, with a particular focus on international humanitarian questions.
The Media Portrayals of Minorities Project at Middlebury College uses digital sources to better understand depictions of minorities in the media. Our goal is to track and explain how and why media representations of groups shift over time, vary across place, or compare to one another.