低资源多语言翻译评测榜单

低资源多语言翻译评测榜单由鹏城实验室和粤港澳大湾区数字经济研究院等单位共同举办，该榜单由Bright多语言翻译平台做技术支持，Bright是一个以中文为中心面向“一带一路”国家低资源语种的机器翻译模型提供评测，该数据集经过资深的语言专家标注、严格的质量管控流程和多轮精细校验，拥有高准确度，并覆盖科技、政治、医疗、旅游、新闻、日常等6大内容领域。旨在致力于打造全球开放、共建共享的交流平台，为提升“一带一路”沿线国家语言同中文之间的互译能力而努力。<br />
官网地址：https://bright.pcl.ac.cn/ <br />
GitHub：https://github.com/LJYmaodou/CCEval

The Low-Resource Multilingual Translation Evaluation Benchmark is co-hosted by Peng Cheng Laboratory, the Institute of Digital Economy of Guangdong-Hong Kong-Macao Greater Bay Area and other institutions. This benchmark is technically supported by the Bright multilingual translation platform, which focuses on Chinese as the core and provides evaluation services for machine translation models targeting low-resource languages in countries along the Belt and Road Initiative. This dataset has been annotated by senior language experts, undergone strict quality control procedures and multiple rounds of meticulous verification, ensuring high accuracy, and covers six major content domains including technology, politics, healthcare, tourism, news and daily scenarios. It aims to build a globally open, jointly-developed and shared exchange platform, and devote itself to improving the mutual translation capabilities between Chinese and the languages of countries along the Belt and Road Initiative.<br />Official website: https://bright.pcl.ac.cn/<br />GitHub: https://github.com/LJYmaodou/CCEval

搜集汇总

背景与挑战

背景概述

该数据集是低资源多语言翻译评测榜单，由鹏城实验室和粤港澳大湾区数字经济研究院等单位联合推出，旨在评估中文与12个低资源语种（如越南语、阿拉伯语等）的机器翻译性能。它覆盖科技、政治等六大领域，包含24个翻译方向，每个方向约1200条平行句对，总计28668条测试数据，采用BLEU指标进行评测。