WDV

WDV是一个由Wikidata构建的大型知识图谱数据口头化数据集，包含7607条记录，旨在将结构化和半结构化数据转换为人类可读格式。数据集紧密结合了三元组和文本，覆盖了广泛的实体和谓词类型。创建过程中，通过众包方式收集了1400条记录的人工标注，用于评估口头化质量。WDV的应用领域包括自动化事实检查和开放领域问答，旨在解决如何有效将知识图谱数据口头化的问题。WDV is a large-scale knowledge graph data verbalization dataset constructed from Wikidata. It contains 7,607 records and aims to convert structured and semi-structured data into human-readable formats. The dataset tightly integrates triples and text, covering a wide range of entity and predicate types. During its development, 1,400 records were manually annotated via crowdsourcing to evaluate the quality of verbalization. Application scenarios of WDV include automated fact-checking and open-domain question answering, aiming to address the problem of how to effectively verbalize knowledge graph data.