Law's Semantic Self-Portrait: Discerning Doctrine with Co-Citation Networks and Keywords
An apex court’s body of cases has an internal texture, continually augmented by recent citations to earlier, topically related cases. How can we best describe that texture? The citation network shows a path. Specifically, what past Supreme Court cases do more recent Supreme Court cases tend to cite together, as if a topical pair? Using a web of those oft-cited pairs, what noun phrases appear in a given cluster of cases more often, relative to the rate at which those phrases appear in writings more generally? To answer these questions is to map, in detail, a body of decisional law. Using common network-analysis and corpus-linguistics tools, one can derive from a group of cases the key empirical facets of the legal doctrine embodied in that cluster of cases — a semantic self-portrait that the cases paint with their own words and citations. This paper provides a pair of case studies for revealing the latent semantic building blocks of legal doctrine. First, using a new citation dataset, I analyze the co-citation network of a sharply defined group of Supreme Court cases (in this instance, cases on the Warsaw Convention, a treaty that limits liability for loss or injury in international air travel, and other cases related thereto). Second, building on a citation dataset from prior work, I analyze the co-citation network of all the Supreme Court’s intellectual property cases from 1947 to 2018, inclusive. With these empirical studies, I show that co-citation analysis complements both traditional legal analysis — by establishing data about legal doctrine, from the bottom up, using large case networks — and the attitudinal-model studies from political science — by focusing on the substance of legal doctrine, rather than on judges’ votes in split cases placed on a right-left continuum.