Knowledge Graph Completion
Today, we are experiencing an unprecedented production of resources, published as Linked Open Data (LOD, for short) and made accessible through the Web of data. This is leading to the creation of knowledge graphs (KGs) containing billions of RDF (Resource Description Framework) triples, such as DBpedia, YAGO and Wikidata on the academic side, and the Google Knowledge Graph, Facebook Graph or Microsoft’s Satori graph on the commercial side.
These KGs contain millions of entities (such as people, proteins, or books), and millions of facts about them. This knowledge is typically expressed in RDF, i.e., as triples of the form ⟨Macron, presidentOf, France⟩. Some KGs provide an ontology expressed in OWL2 (Web Ontology Language), which describes the vocabulary (the classes and properties) for the RDF facts. However, these KGs are big, incomplete by nature and may contain errors and contradictions.
Therefore, to be able to enrich and expand such knowledge graphs several problems have to be dealt like information extraction, ontology matching, identity link detection/invalidation and knowledge discovery. Knowledge graph completion is an active research area, which has recently witnessed impressive interest in both academia side and industry side.
In this tutorial we will give an overview of existing approaches for identity link detection and invalidation, and for key discovery, in a way that is accessible to researchers in Web technologies area.