Go to ACL ARR 2025 February homepageBridging Legal Language Gaps in Hong Kong: A Multi-Agent Framework for Context-Aware Translation of Judgments
Abstract: Multi-agent systems empowered by large language models (LLMs) have demonstrated remarkable capabilities in a wide range of downstream applications, including machine translation. However, translating Hong Kong legal judgments remains an exceptionally challenging task due to its intricate legal lexicon, culturally embedded nuances, and complex linguistic structures. In this work, we introduce \textsc{Tapagents}, a novel multi-agent translation system inspired by real-world case law translation workflow. \textsc{Tapagents} employs specialized agents — Translator, Annotator, and Proofreader
— to collaboratively produce translations that are Accuracy in Legal Meaning, Appropriateness in Style, and Coherence and Cohesion in Structure. Our system supports customizable LLM configurations and achieves 3,972× cost reduction compared to professional human services. Evaluations show \textsc{Tapagents} surpasses ChatGPT-4o in legal semantic accuracy, structural coherence, and stylistic fidelity, yet trails human experts in contextualizing complex terminology and stylistic naturalness.Our live demo website is available at \footnote{\url{}}. Our demonstration video is available at \footnote{\url{}}.
Paper Type: Long
Research Area: Machine Translation
Research Area Keywords: Machine Learning for NLP
Contribution Types: NLP engineering experiment, Data resources
Languages Studied: English and Chinese
Submission Number: 8094
Loading