Title: OntoLegal Models and LLMs for Harmonizing Multilingual Legal Terms in Comparative Jurisprudence
Authors: Khujakulov Sunnatullo
Volume: 9
Issue: 11
Pages: 53-62
Publication Date: 2025/11/28
Abstract:
The globalization of legal discourse demands advanced strategies for achieving semantic interoperability among diverse legal systems. This study introduces a hybrid framework that integrates OntoLegal Models and Large Language Models (LLMs) to harmonize multilingual legal terminology in comparative jurisprudence. OntoLegal Models provide a structured representation of legal concepts, hierarchies, and interrelations across jurisdictions, while LLMs, trained on multilingual legal corpora, are utilized for context-sensitive extraction, clustering, and disambiguation of legal terms. The research applies this integrated approach to English, Uzbek, and Russian legal datasets, focusing on terminological equivalence within constitutional, criminal, and financial law domains. Results demonstrate that the OntoLegal-LLM framework achieves an average semantic alignment accuracy of 89%, outperforming traditional ontology-only methods by 17%. Cluster analyses reveal that LLM-based embeddings successfully capture functional equivalences where direct lexical equivalents are absent, particularly in culturally embedded legal concepts. Moreover, the integration reduces terminological inconsistencies across translations by 23%, contributing to greater clarity and legal interoperability. The findings confirm that combining ontological modeling with AI-driven linguistic analysis enhances the precision and standardization of legal terminology across jurisdictions. This research advances the disciplines of jurilinguistics, computational legal linguistics, and comparative law, providing a scalable model for developing multilingual legal knowledge graphs, improving legal translation accuracy, and supporting future cross-border digital legislation initiatives.