NOWJ@COLIEE 2024: Leveraging Advanced Deep Learning Techniques for Efficient and Effective Legal Information Processing

Tan-Minh Nguyen, Hai-Long Nguyen, Dieu-Quynh Nguyen, Hoang-Trung Nguyen, Thi-Hai-Yen Vuong, Ha-Thanh Nguyen

Abstract: The constantly expanding volume of legal information presents a growing challenge for legal professionals to efficiently handle their workload. COLIEE is an annual competition organized with four tasks about automated legal information processing, which aims to liberate some of the pressures. This paper presents our approaches for all four tasks in the COLIEE 2024, which involve legal case retrieval (task 1), legal case entailment (task 2), statute law retrieval (task 3), and legal textual entailment (task 4). We utilize state-of-the-art deep learning models like BERT, Longformer, large language models, and a joint learning approach. Of a total of 18 submissions in task 2, our best run was ranked sixth on the leaderboard. For tasks 3 and 4, our submissions achieved fourth and fifth places out of nine teams in the overall ranking.