Keywords: Web of Data, Knowledge Graph Refinement, Acyclic Graphs
Abstract: The publication of knowledge graphs on the Web in the form of RDF datasets, and the subsequent combination of such knowledge graphs are both essential to the idea of Linked Open Data. Combining such knowledge graphs can result in undesirable graph structures and even in logical inconsistencies. Refinement methods that can detect and repair such undesirable graph structures are therefore of crucial importance. Existing refinement methods for knowledge graphs are often domain-specific, are limited to single relations (e.g. owl:sameAs), or are limited in scale. To tackle this new challenge, we present a challenge consisting of a number of datasets of transitive and pseudo-transitive relations together with hand-annotated gold standards, and a set of baseline algorithms. We introduce an efficient Web-scale knowledge graph refinement algorithm that works for such relations. Our algorithm analyses the graph structure, and allows the use of a weighting scheme to heuristically determine which possibly erroneous links should be removed to make the graph cycle free. When compared against general-purpose graph algorithms that perform the same task, our algorithm removes the least amount of links to make the graph of transitive relations cycle-free while maintaining a better precision in identifying erroneous links as measured against a human gold-standard.
First Author Is Student: Yes
Subtrack: Data Dynamics, Quality, and Trust