MuST-SHE
收藏资源简介:
MuST-SHE数据集由特伦托大学创建,是一个多语种的自然语言处理基准数据集,专注于评估性别偏见。该数据集包含约2136个(音频, 转录, 翻译)三元组,适用于英语到意大利语和法语的翻译方向。每个三元组都经过精心标注,以反映不同类型的性别现象。数据集的创建过程涉及从原始数据中筛选和手动检查,确保数据的质量和性别现象的平衡分布。MuST-SHE数据集的应用领域主要集中在评估和改进机器翻译系统在处理性别相关内容时的性能,旨在解决性别偏见问题,提高翻译的准确性和公正性。
The MuST-SHE dataset, developed by the University of Trento, is a multilingual natural language processing (NLP) benchmark dataset dedicated to evaluating gender bias. It contains approximately 2,136 (audio, transcription, translation) triplets, supporting English-to-Italian and English-to-French translation tasks. Each triplet is meticulously annotated to capture a wide range of gender-related phenomena. The dataset construction process involves screening raw data and conducting manual inspections to ensure data quality and a balanced distribution of gender-related content. The primary applications of the MuST-SHE dataset focus on evaluating and enhancing the performance of machine translation systems when handling gender-related content, aiming to address gender bias issues and improve the accuracy and fairness of machine translation.




