Knowledge Graphs (KG) have gained increasing importance in science, business and society in the last years. However, most knowledge graphs were either extracted or compiled from existing sources. There are only relatively few examples where knowledge graphs were genuinely created by an intertwined human-machine collaboration. Also, since the quality of data and knowledge graphs is of paramount importance, a number of data quality assessment models have been proposed. However, they do not take the specific aspects of intertwined human-machine curated knowledge graphs into account. In this work, we propose a graded maturity model for scholarly knowledge graphs (KGMM), which specifically focuses on aspects related to the joint, evolutionary curation of knowledge graphs for digital libraries. Our model comprises 5 maturity stages with 20 quality measures. We demonstrate the implementation of our model in a large scale scholarly knowledge graph curation effort.
翻译:过去几年来,知识图(KG)在科学、商业和社会中越来越重要,然而,大多数知识图或是从现有来源提取或汇编的,只有相对较少的例子表明知识图真正是由相互交织的人类机器合作产生的,此外,由于数据和知识图的质量至关重要,因此提出了若干数据质量评估模型,但没有考虑到相互交织的人类机器知识图的具体方面。在这项工作中,我们建议为学术知识图(KGMM)提供一个分级成熟模型,该模型特别侧重于数字图书馆知识图的联合、进化整理方面。我们的模型包括五个成熟阶段,有20种高质量的措施。我们展示了我们模型在大规模学术知识图整理工作中的实施情况。