LegalViz: Legal Text Visualization by Text To Diagram Generation
Abstract: Legal documents including judgments, court orders, government ordinances, professional papers, and textbooks of judicial examinations require highly sophisticated legal knowledge for understanding. To disclose expert knowledge for non-experts, we explore the problem of visualizing legal texts with easy-to-understand diagrams and propose a novel dataset of LegalViz with 23 languages and 5,580 cases of legal document and visualization pairs, using the DOT graph description language of Graphviz. LegalViz provides a simple diagram from a complicated legal corpus identifying legal entities, rules, statements, and transactions at a glance, that are important in each judgment. In addition, we provide a new evaluation approach for the legal diagram visualization by considering the graph and text similarities. We conducted empirical studies on few-shot and finetuning large language models for generating legal diagrams and evaluated them with the graph and text evaluation metrics by each model in 23 languages and confirmed the effectiveness of our dataset.
Paper Type: Long
Research Area: NLP Applications
Research Area Keywords: Legal NLP, Text to Graph
Contribution Types: Data resources
Languages Studied: Bulgarian, Spanish, Czech, Danish, German, Estonian, Greek, English, French, Croatian, Italian, Latvian, Lithuanian, Hungarian, Maltese, Dutch, Polish, Portuguese, Romanian, Slovak, Slovenian, Finnish, Swedish
Submission Number: 5378
Loading