Dynamic Relation Repairing for Knowledge Enhancement

Dynamic relation repair aims to efficiently validate and the instances for knowledge graph enhancement (KGE), where KGE captures missing relations from unstructured data leads noisy facts graph. With prosperity of data, an online approach is asked clean new RDF tuples before adding them base. To tuples, constraint processing a common but intractable approach. Plus, when graph, patterns would be created, whereas explicit discovery constraints also intractable. Therefore, although dynamic has unfortunate hardness, it necessary enhancing graphs effectively under fast-growing data. Motivated by this, we establish repairing structure analyze its hardness on basic operations. ensure validation, introduce implicit constraints, approximate matching, linkage prediction based localized patterns. efficiently, further study cold start problems processing. Experimental results real datasets demonstrate that our proposed can capture with wrong labels dynamically effectively.