Understanding legal differences between Japan and foreign countries is essential but challenging due to complex legal texts. This study develops an automated system for mapping similar provisions using a multilingual model and fine-tuning with a legal corpus. Unlike previous one-to-one mapping methods, our approach supports one-to-many mappings. Experiments on Japanese and German Civil Codes showed improved accuracy, leveraging a government-provided translation to overcome language barriers. Fine-tuning further improved performance, demonstrating our approach’s effectiveness. Our method contributes to comparative law research by enabling more accurate and efficient legal provision mapping across jurisdictions.