SIRIS-Lab/affilgood-contrastive-dataset

AffilGood对比数据集是专为训练识别和链接组织到它们官方表示的实体链接模型而设计的数据集。它包含52,900个由AffilGood框架生成的组织查询样本和各种表示形式的负例，基于研究组织注册簿（ROR）数据。该数据集格式化为支持对比学习任务，并可用于训练模型。

The AffilGood Contrastive Dataset is designed for training entity linking models, specifically for identifying and linking organizations to their official representations. It includes 52,900 samples of organization queries with various representations and negative examples, generated using the AffilGood framework, based on the Research Organization Registry (ROR) data. The dataset is formatted to support contrastive learning tasks and can be used to train models.