This year was the 10th anniversary of the Competition on Legal Information Extraction and Entailment (COLIEE 2024). The objective of the competition is to build state of art legal information retrieval and entailment systems. This year, the COLIEE competition included four challenges concerning a combination of case law and statute law. Tasks included finding supporting cases, finding relevant paragraphs from previous cases, considering yes/no legal questions, and retrieving statutes from a database.
Our involvement was a natural fit, as the competition's theme beautifully aligns with our lab's focus, said Animesh Nighojkar, Ph.D. Student. The Advancing Machine and Human Reasoning Lab (AMHR), led by Assistant Professor John Licato, is a cross-disciplinary lab dedicated to improving AIs cognitive-level reasoning processes such that it improves and facilitates, rather than replaces, human reasoning.
The competition consisted of four tasks. The AMHR team submitted three approaches to three of the tasks, and won the top prize on one of them, placing competitively in the other two. Competitions like this are a rollercoaster of emotions. You put your heart and soul into your work, yet the outcome is always unknown, especially with a hidden test set, said Animesh. AMHRs winning approach leveraged a fusion of domain-specific knowledge and a cutting-edge text ranking model, monoT5, tailored meticulously for pinpoint legal case analysis. Their work could transform weeks of exhaustive research into a few minutes of insightful discovery.
Its about enhancing human capabilities, not replacing them. Our developed technique is a game-changer for legal professionals. It empowers them to rapidly identify pertinent precedent cases, significantly aiding in various legal processes like motions, negotiations, and trials.
For more information on COLIEE 2024,
For more information on AMHR, see their