FrancophonIA/Human-reviewed_automatic_English_translations_Europeana

该数据集包含了来自Europeana平台元数据的人类审核或后期编辑的翻译。翻译涉及从17种欧洲语言到英语的转换。其中，来自保加利亚语、克罗地亚语、捷克语、丹麦语、德语、希腊语、西班牙语、芬兰语、匈牙利语、波兰语、罗马尼亚语、斯洛伐克语、斯洛文尼亚语和瑞典语的翻译由语言专家小组（每种语言2名专家）审核。来自荷兰语、法语和意大利语的翻译来源于Europeana收藏中的时尚、视听和博物馆遗产领域，并由文化遗产专家评估。评估者被要求对自动翻译在0到100的范围内进行评分。文本片段是从Europeana数据模型的不同元数据属性中提取的，这些属性捕获了文化遗产项目的各个方面，例如一幅画的标题或描述。TSV文件包含了获得90%或以上人类评分的自动翻译。对于意大利语、法语和荷兰语，还包括了经过人类专家后期编辑的自动翻译，以反映正确的翻译。TTSV文件的第一行是源语言，第二行是英语。

The dataset includes human-reviewed or post-edited translations of metadata sourced from the Europeana platform. The translations are from 17 European languages to English. Translations from Bulgarian, Croatian, Czech, Danish, German, Greek, Spanish, Finnish, Hungarian, Polish, Romanian, Slovak, Slovenian, and Swedish have been reviewed by a group of linguist experts (2 experts for each language). Translations from Dutch, French, and Italian are sourced from Europeana collections in the fashion, audiovisual, and museum heritage domains and have been evaluated by cultural heritage experts. Evaluators were asked to rate the automatic translations on a scale from 0 to 100. The textual segments were extracted from different metadata properties of the Europeana Data Model, which captures aspects of a CH item, such as the title of a painting or its description. The TSV files include automatic translations that received a human rating of 90% or above. For Italian, French, and Dutch, post-edited automatic translations by human experts are also included in the TSV files to reflect a correct translation. TTSV files have the first row in the source language and the second in English.