Constructing and Cleaning Identity Graphs in the LOD Cloud

• Published in: Data intelligence Vol. 2; no. 3; pp. 323 - 352
• Main Authors: Raad, Joe, Beek, Wouter, van Harmelen, Frank, Wielemaker, Jan, Pernelle, Nathalie, Saïs, Fatiha
• Format: Journal Article
• Language: English
• Published: One Rogers Street, Cambridge, MA 02142-1209, USA MIT Press 01.07.2020; MIT Press Journals, The
• Subjects: Graph theory; Identity; Linked Open Data; Names; Quality; Reasoning; Semantic web; Semantics
• ISSN: 2641-435X; 2641-435X
• DOI: 10.1162/dint_a_00057

In the absence of a central naming authority on the Semantic Web, it is common for different data sets to refer to the same thing by different names. Whenever multiple names are used to denote the same thing, owl:sameAs statements are needed in order to link the data and foster reuse. Studies that date back as far as 2009, observed that the owl:sameAs property is sometimes used incorrectly. In our previous work, we presented an identity graph containing over 500 million explicit and 35 billion implied owl:sameAs statements, and presented a scalable approach for automatically calculating an error degree for each identity statement. In this paper, we generate subgraphs of the overall identity graph that correspond to certain error degrees. We show that even though the Semantic Web contains many erroneous owl:sameAs statements, it is still possible to use Semantic Web data while at the same time minimising the adverse effects of misusing owl:sameAs.