On the Use and Abuse of Word Embeddings for Digital Humanities

Speaker: Christof Schöch, Professor of Digital Humanities at the University of Trier

This talk will provide an introduction to Word Embeddings in the Digital Humanities, particularly in Computational Literary Studies and Digital History. Word Embeddings represent meaning in a high-dimensional vector space and therefore make semantic meaning amenable to calculations and measures. They promise to bring us closer to machines that understand literary texts or historical documents in context. More concretely, they are a highly useful building-block in computational approaches to such texts, either as enhancements for existing computational approaches (such as Topic Modeling or Sentiment Analysis), or as fundamentally new evidence (for example in the domain of history of ideas). This talk will present some of the avenues opened up by Word Embeddings for research in the Digital Humanities