WikiSRS

该数据集是由亚马逊土耳其机器人工作人员为配对的维基百科实体分配相似性和相关性判断的新颖数据集。其设计流程遵循了Pakhomov等人（2010年）的方法，为在更广泛背景下评估实体嵌入提供了一个框架。该数据集包含了688对相似性和相关性判断，其任务是对维基百科实体嵌入进行评估。

This novel dataset is constructed by having Amazon Mechanical Turk workers assign similarity and relevance judgments to paired Wikipedia entities. Its design follows the methodology proposed by Pakhomov et al. (2010), providing a framework for evaluating entity embeddings in a broader contextual scope. The dataset contains 688 pairs of similarity and relevance judgments, which are tailored for the task of assessing Wikipedia entity embeddings.