多语种法律翻译数据集

本数据集面向法律翻译质量评估，基于覆盖多司法管辖区与多语言的公开数据集Fairlex构建。数据集选取其中的英语、德语及法语原句，构建了包含6000余条高质量平行语料的数据资源。语料涵盖了复杂的法律语言现象与典型翻译难点，能够支持不同法律领域、难度级别及语言对的翻译测试与研究，为法律翻译的评估、教学与实践提供了多样化的场景与重要参考。 本数据集聚焦于法律领域的翻译质量评估，在数据集的构建过程中，我们特别选用了法律领域公开数据集Fairlex作为基础资源。Fairlex数据集具有广泛的覆盖范围，涵盖了四个具有代表性的司法管辖区，即欧洲理事会、美国、瑞士和中华人民共和国的法律文本。同时，它还涉及五种语言，包括英语、德语、法语、意大利语和中文，为法律翻译评估提供了丰富的语言素材。Fairlex数据集包含四个子数据集，每个子数据集都针对特定的司法管辖区和语言环境进行了精心设计。这种设计确保了评估的广泛性和多样性，使得我们能够从不同角度、不同层面深入评估法律翻译的特点和难点。本数据集则基于Fairlex中的英语、德语、法语原句构建而成，这些原句不仅具有高度的代表性，能够充分反映法律文本的语言特征和翻译需求，还囊括了法律文本中常见的复杂语言现象和潜在的翻译难点，如专业术语的准确翻译、句法结构的灵活转换以及文化背景的处理等。该数据集共计收集了6000余条高质量的平行语料，这些语料涵盖了不同法律领域、不同难度级别和不同语言对，为法律翻译测试提供了丰富的素材和多样的场景。无论是对于法律翻译研究、法律翻译教学还是法律翻译实践，本数据集都具有重要的参考价值和实用意义。This dataset is developed for legal translation quality assessment, and is constructed based on the publicly available multilingual and multi-jurisdictional Fairlex dataset. We selected original sentences in English, German and French from Fairlex to build a resource containing over 6,000 high-quality parallel corpora. The corpora cover complex legal linguistic phenomena and typical translation challenges, enabling translation testing and research across different legal domains, difficulty levels and language pairs, and providing diversified scenarios and important references for legal translation evaluation, teaching and practice. This dataset focuses on legal translation quality assessment, and we specifically adopted the publicly available legal dataset Fairlex as the foundational resource during its construction. The Fairlex dataset has extensive coverage, encompassing legal texts from four representative jurisdictions: the Council of Europe, the United States, Switzerland, and the People's Republic of China. It also covers five languages, including English, German, French, Italian and Chinese, providing abundant linguistic materials for legal translation evaluation. The Fairlex dataset consists of four sub-datasets, each meticulously designed for specific jurisdictions and linguistic contexts. This design ensures the breadth and diversity of the evaluation, allowing us to thoroughly assess the characteristics and challenges of legal translation from multiple perspectives and dimensions. This dataset is built upon the original English, German and French sentences from Fairlex. These original sentences are not only highly representative, fully reflecting the linguistic features and translation requirements of legal texts, but also cover common complex linguistic phenomena and potential translation challenges in legal texts, such as accurate translation of professional terminology, flexible conversion of syntactic structures, and handling of cultural backgrounds, among others. A total of over 6,000 high-quality parallel corpora have been collected in this dataset, which cover different legal domains, difficulty levels and language pairs, providing abundant materials and diverse scenarios for legal translation testing. This dataset holds important reference value and practical significance for legal translation research, teaching and practice alike.

搜集汇总

背景与挑战

背景概述

该数据集基于多司法管辖区与多语言的Fairlex公开数据集构建，选取英语、德语和法语原句，形成了包含6000余条高质量平行语料的数据资源。它涵盖了法律领域的复杂语言现象和典型翻译难点，旨在支持法律翻译质量评估、教学与研究，提供多样化的测试场景和重要参考。