DS 5559 Exploratory Text Analytics

Exploratory text analytics concerns the application of computational and statistical methods to the interpretation of large collections of digitized written documents. The field is motivated by the research of scholars from the humanities and human sciences interested in understanding the semantic, cultural, and social dimensions of texts from historical and contemporary sources, such as novels, newspapers, and social media. The course comprises three main sections: (1) an overview of text interpretation theory combined with information theory to introduce the domain knowledge required for making inferences in this area; (2) a hands-on introduction to methods for converting unstructured textual content into both graph and vector-space representations; and (3) the application and discussion of algorithms from natural language processing and text mining approaches, including term frequency measures, topic models, and sentiment analyses, to address the classic problems of text classification and clustering as well as new areas, such as social event detection, computational narratology, and data-driven approaches to structuralist poetics. There are no hard prerequisites for the course, but students should be comfortable with combining qualitative and quantitative approaches, have some experience programming in Python, and be familiar with basic statistics and probability theory.