PRILJ: an efficient two-step method based on embedding and clustering for the identification of regularities in legal case judgments

Abstract In an era characterized by fast technological progress that introduces new unpredictable scenarios every day, working in the law field may appear very difficult, if not supported right tools. this respect, some systems based on Artificial Intelligence methods have been proposed literature, to support several tasks legal sector. Following line of research, paper we propose a novel method, called PRILJ, identifies paragraph regularities case judgments, experts during redaction documents. Methodologically, PRILJ adopts two-step approach first groups documents into clusters, according their semantic content, and then paragraphs for each cluster. Embedding-based are adopted properly represent numerical feature space, Approximated Nearest Neighbor Search method is efficiently retrieve most similar with respect document under preparation. Our extensive experimental evaluation, performed real-world dataset provided EUR-Lex, proves effectiveness efficiency method. particular, its ability modeling different topics documents, as well capturing semantics textual beneficial considered task, make robust possible presence noise data.