LegalViz: Legal Text Visualization by Text To Diagram Generation
Proceedings of the 2025 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers) · June 2025
Novel approach for visualizing complex legal document structures through diagram generation from text.
BibTeX
@inproceedings{onami2024legalviz,
title = {LegalViz: Legal Text Visualization by Text To Diagram Generation},
author = {Eri Onami and Taiki Miyanishi and Koki Maeda and Shuhei Kurita},
booktitle = {Proceedings of the 2025 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)},
year = {2025},
address = {Albuquerque, USA},
publisher = {Association for Computational Linguistics}
}Abstract
LegalViz is a novel dataset and method designed to visualize complex legal documents as intuitive diagrams using Graphviz’s DOT language. The dataset comprises 7,010 pairs of professional legal texts and corresponding diagrams, covering 23 languages from EUR-LEX. LegalViz facilitates understanding of legal documents by highlighting essential entities, relationships, transactions, and underlying legal rules. Experimental results demonstrate that large language models fine-tuned with LegalViz significantly outperform traditional methods, validating the effectiveness of our approach.