Scholarlink

The analysis of legal documents, specifically Indian ones, which are characterized by their length and terminological complexity, poses significant challenges, as traditional approaches are not adequate for drawing conclusions efficiently. This article presents an evaluation of neural network models for predicting rhetorical roles in Indian court judgments, as part of Subtask A of Task 6 of SemEval-2023, dedicated to the understanding of legal texts. Experiments were conducted with models such as BERT, LegalBERT and RoBERTa to classify Rhetorical Roles (RR) in Indian legal texts. The results show that the LegalBERT model performed best, with an F1 score of 0.6318 (63.18%) compared to the other models tested.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Comparison of Neural Network Models for Predicting Rhetorical Roles in Legal Documents

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Hansel Lázaro Valdés Pérez,
  • Gabriela Giovana Pérez López,
  • Darnes Vilariño Ayala,
  • Beatriz Beltrán Martínez,
  • David Pinto Avendaño,
  • Carmen Cerón Garnica

摘要

The analysis of legal documents, specifically Indian ones, which are characterized by their length and terminological complexity, poses significant challenges, as traditional approaches are not adequate for drawing conclusions efficiently. This article presents an evaluation of neural network models for predicting rhetorical roles in Indian court judgments, as part of Subtask A of Task 6 of SemEval-2023, dedicated to the understanding of legal texts. Experiments were conducted with models such as BERT, LegalBERT and RoBERTa to classify Rhetorical Roles (RR) in Indian legal texts. The results show that the LegalBERT model performed best, with an F1 score of 0.6318 (63.18%) compared to the other models tested.