Deriving Insights from Legal Data

"Deriving Insights from Legal Data" is an episode in the IdeaFlow series hosted by featuring discussions on computational insights from legal data. The event is organized by Dazza Greenwood, Executive Director of and features guest speaker Robert Mahari, a joint JD/PhD candidate at Harvard Law School and the MIT Media Lab’s Human Dynamics Group.

The episode covers two main topics. First, it examines the law as a knowledge system that evolves through citations, comparing citation networks in law and science using tools from the "science-of-science" field. This analysis reveals that, despite fundamental differences, the core citation dynamics in both domains are remarkably universal, suggesting they are shaped by intrinsic human constraints. Second, the discussion explores how legal citation data can be utilized to build advanced natural language processing models. Specifically, it highlights a state-of-the-art BERT model trained on 530,000 examples of legal arguments made by U.S. federal judges, which predicts relevant passages from precedential court decisions given a brief legal argument, achieving a top-10 prediction accuracy of 96% on unseen examples.

The event also includes a segment on the upcoming special release of the MIT Computational Law Report on "Composable Governance," providing attendees with information on how to get involved in this initiative.

This episode is particularly beneficial for legal professionals, data scientists, and researchers interested in the intersection of law and computational methods. Attendees will gain insights into the application of machine learning and network science to legal data, enhancing their understanding of how computational tools can improve legal processes and access to justice.