Discovering Relationships Among Properties in Wikidata Knowledge Graph

Wikidata, a community-maintained knowledge graph (KG), may integrate different entities and properties with the same meaning. Contributors can add new properties similar in meaning to other properties in the KG. Detecting relationships among these properties plays a crucial role in interoperability and downstream tasks. We tackle the problem of discovering relationships among properties regarding their instances in Wikidata. Although our approach is knowledge graph-agnostic, it can be applied to any KGs. Our approach resorts to Class-based Relationship Discovery (CRD) to capture the most important characteristics of the properties. We evaluate our approach over Wikidata and show the benefits of exploiting statements annotated with qualifiers, references, and ranks. We empirically study the distribution and frequency of relationships among predicates in six domains to provide evidence that KGs enclose relationships that define the same real-world properties.