gaHealth

gaHealth是由芒斯特理工大学和都柏林城市大学的研究人员开发的首个英语-爱尔兰语双语健康数据集。该数据集包含16,201条平行文本，主要来源于爱尔兰政府发布的专业翻译文档，如健康部门的战略声明和年度报告。创建过程中，研究人员采用了一套工具链来提取、清洗和选择合适的句子以构建双语语料库。gaHealth数据集专为解决低资源语言在健康领域的机器翻译问题而设计，通过实验验证，使用该数据集训练的模型在BLEU评分上取得了显著提升，特别是在处理健康相关数据时。gaHealth is the first English-Irish bilingual health dataset developed by researchers from Munster Technological University and Dublin City University. The dataset contains 16,201 parallel text pairs, mainly sourced from professionally translated documents released by the Irish government, such as health sector strategic statements and annual reports. During its development, the researchers adopted a toolchain to extract, clean and select appropriate sentences for the construction of this bilingual corpus. The gaHealth dataset is specifically designed to address machine translation challenges for low-resource languages in the healthcare domain. Experimental results have verified that models trained on this dataset achieve significant improvements in BLEU scores, especially when processing healthcare-related data.