Computer Science and Engineering

MIDAS Seminar

Word Embeddings: What Works, What Doesn’t, and How to Tell the Difference for Applied Research

Arthur SpirlingDeputy Director,Center for Data Science, New York University
340 West HallMap

Arthur considers the properties and performance of word embeddings techniques in the context of political science research. In particular, his team explores key parameter choices with respect to efficiency and quality of inferences possible with these models. Reassuringly he will show that results are generally robust to such choices for political corpora of various sizes and in various languages. For completeness, he provides best practice advice for cases where local fitting is required.