Abstract
This Article applies computational legal methods to antitrust caselaw. We introduce modifications to topic modeling that streamline big data legal analysis. Topic modeling, an algorithm that maps the statistical relationships among words, is quickly gaining traction in law. Yet its adherents have overlooked criticisms of the tool from computer science and digital humanities, especially around decontextualization.
We utilize aggregated modeling, an improvement over traditional topic models, to analyze two corpora of antitrust decisions extracted from the Caselaw Access Project. One corpus covers market power cases, while the other features the terms ‘antitrust’ and ‘regulation.’
Though technical advances have made the extraction of legal data much easier, the development of analytical tools—especially for researchers without statistical training—lags markedly. Our aggregated modeling provides greater contextualization of terms and topics. In publicizing our code and methodology, we hope that our results will prompt legal research providers to be more forthcoming with their search algorithms, which are notoriously opaque.
Recommended Citation
Felix B. Chang, Erin McCabe & James Lee,
Modeling the Caselaw Access Project: Lessons for Market Power and the Antitrust–Regulation Balance,
22
Nev. L.J.
685
(2022).
Available at:
https://scholars.law.unlv.edu/nlj/vol22/iss2/7