Knowledge graphs and ontologies are becoming increasingly important as technical solutions for Findable, Accessible, Interoperable, and Reusable data and metadata (FAIR Guiding Principles). We discuss four challenges that impede the use of FAIR knowledge graphs and propose semantic units as their potential solution. Semantic units structure a knowledge graph into identifiable and semantically meaningful subgraphs. Each unit is represented by its own resource, instantiates a corresponding semantic unit class, and can be implemented as a FAIR Digital Object and a nanopublication in RDF/OWL and property graphs. We distinguish statement and compound units as basic categories of semantic units. Statement units represent smallest, independent propositions that are semantically meaningful for a human reader. They consist of one or more triples and mathematically partition a knowledge graph. We distinguish assertional, contingent (prototypical), and universal statement units as basic types of statement units and propose representational schemes and formal semantics for them (including for absence statements, negations, and cardinality restrictions) that do not involve blank nodes and that translate back to OWL. Compound units, on the other hand, represent semantically meaningful collections of semantic units and we distinguish various types of compound units, representing different levels of representational granularity, different types of granularity trees, and different frames of reference. Semantic units support making statements about statements, can be used for graph-alignment, subgraph-matching, knowledge graph profiling, and for managing access restrictions to sensitive data. Organizing the graph into semantic units supports the separation of ontological, diagnostic (i.e., referential), and discursive information, and it also supports the differentiation of multiple frames of reference.
翻译:作为可查找、可获取、可互操作和可再使用的数据和元元数据的技术解决方案,知识图和知识图越来越重要。我们讨论了阻碍使用FAIR知识图的四项挑战,并提议将语义单位作为潜在的解决方案提出。语义单位将知识图组成为可识别和具有语义意义的子图。每个单位都以其本身的资源为代表,即时将相应的语义单位分类作为相应的语义单位类别,并可以作为FAIR数字目标以及RDF/OWL和属性图表中的纳米发布方式加以实施。我们将语义和复合单位作为语义图表的基本类别加以区分。语义单位代表最小、独立、对读者具有语义意义的语义图表。我们将语义、特质、通用语义单位作为语义单位的基本类型,并提议其代表性和正式的语义体系体系(包括缺席语言、读取、基本信息限制),不包含空白的语义和复合语义的语义说明,同时将多种语义的语义表解结构单位转化为。