Optimizing Keyphrase Extraction for Court Decisions Using Legal References
Zoltán Szoplák, Peter Gurský, Dávid Varga · Frontiers in artificial intelligence and applications · 2024
This paper presents preliminary methods and results on extracting keyphrases from Slovak Supreme Court decisions to be used in the future by a recommender system for expert annotators employed by the court. We developed methods using referenced law articles and court decisions to improve traditional statistical and semantic approaches for keyphrase extraction. These methods were evaluated using a manually annotated dataset, demonstrating that incorporating extracted references enhances performance, achieving an F1 score percentage of 86.35%.