GAP

GAP数据集是由谷歌AI语言团队创建的，旨在解决自然语言理解中的模糊代词解析问题。该数据集包含8908个从维基百科中抽取的模糊代词-名称对，通过多阶段过滤过程确保数据质量和多样性。GAP数据集特别关注性别公平性，旨在奖励那些性别中立的系统。此数据集的应用领域包括提高下游模型的公平性，如知识库填充中的三元组提取。

The GAP dataset was created by the Google AI Language Team to address the problem of ambiguous pronoun resolution in natural language understanding. It contains 8,908 ambiguous pronoun-name pairs extracted from Wikipedia, and ensures data quality and diversity through a multi-stage filtering process. The GAP dataset pays special attention to gender fairness, aiming to reward gender-neutral systems. Its application areas include improving the fairness of downstream models, such as triple extraction in knowledge base population.