ZeroKBC

ZeroKBC是由腾讯AI实验室创建的综合性零样本知识库补全基准数据集，包含93003条数据，涵盖语言、世界和常识三种知识资源。该数据集旨在解决知识图谱中未见实体和关系的补全问题，通过系统分类和分析，揭示了多个重要的零样本KBC设置。ZeroKBC的应用领域包括问答、常识推理和机器阅读理解，旨在通过零样本学习提升知识图谱的完整性和实用性。

ZeroKBC is a comprehensive zero-shot knowledge base completion (KBC) benchmark dataset developed by Tencent AI Lab. It contains 93,003 data entries, covering three categories of knowledge resources: linguistic knowledge, world knowledge and commonsense knowledge. This dataset aims to solve the completion problem of unseen entities and relations in knowledge graphs, and reveals several important zero-shot KBC settings through systematic classification and analysis. Its application fields include question answering, commonsense reasoning and machine reading comprehension, with the goal of enhancing the completeness and practicality of knowledge graphs via zero-shot learning.