Natural Language Processing for Lawyers and Judges

Abstract

This Essay reviews Michael Livermore and Daniel Rockmore’s edited collection, Law as Data. It discusses each of the chapters, spends some time addressing the differences between predictive and causal inferences for law — an important theme that runs throughout the book — and then turns to a discussion of how natural language processing can help describe legal rules. Contemporary studies of black-letter law which populate today’s treatises and law reviews often rely on cases that have been carefully selected by jurists. As a consequence, distilled statements of law suffer from selection bias regardless of a jurist’s best efforts. Natural language processing, which can describe legal doctrine by examining thousands of cases at once, can help reduce that bias. It can increase confidence in long-standing rules, uncover hidden rationales for their application, and clarify that some matters, such as those embodied in good legal standards, remain best unresolved.

Keywords: Natural Language Processing, Law, Predictive Inference, Causal Inference, Lawyering, Judging