Natural Language Processing for Innovating Behavioral Political Science Research

Abstract

Since the invention of Word2Vec by a Google team in 2013, natural language processing (NLP) techniques have been increasingly applied in the private sector, by government agencies across countries, and in the social sciences. This chapter explains NLP’s basic analytical procedure from preprocessing of raw text data to statistical modeling, reviews the most recent advances in NLP applications in political science, and proposes a new scaling approach for measuring political actors’ spatial preferences along with potential application in decision-making research. It argues that with a greater focus on explaining behavioral mechanisms and processes, which is a goal shared by artificial intelligence/computational modeling and cognitive science, NLP can help improve behavioral political science by its ability to integrate micro-, meso-, and macro-level analyses. Critical and reflexive use of NLP techniques, combined with big data, will lead to obtain better insights on political behavior in general.