FrancophonIA/multilingual_legal_corpus

该数据集是为构建质量估算和自动后编辑模型而创建的训练数据，包含源句子、机器翻译输出、人工后编辑和独立参考翻译的四元组。数据集覆盖了法律领域的在线争议解决、采购和司法等子领域，支持英语与荷兰语、法语、葡萄牙语三种欧洲语言之间的翻译。数据集包含的数据元组数量分别为：英语-荷兰语11249个，英语-法语9989个，英语-葡萄牙语10165个。数据通过神经网络机器翻译系统生成，并进行了匿名化和随机化处理。

This dataset is training data created for constructing quality estimation and automatic post-edition models, containing tuples of source sentences, machine translation outputs, manual post-editions, and independent reference translations. The dataset covers the legal domain, including subdomains such as online dispute resolution, procurement, and justice, and supports translation between English and three European languages: Dutch, French, and Portuguese. The number of data tuples in the dataset is: 11249 for English-Dutch, 9989 for English-French, and 10165 for English-Portuguese. The data were generated using neural network machine translation systems and have been anonymized and randomized.