Network-based Knowledge Graph Assessment

Jan Rörden, Artem Revenko, Bernhard Haslhofer, Andreas Blumauer: Network-based Knowledge Graph Assessment.
In: J. Fernández, S. Hellmann (eds.): Proceedings of the Posters and Demos Track of the 13th International Conference on Semantic Systems - SEMANTiCS2017,
Amsterdam, The Netherlands, September 11-14, 2017,
online: http://ceur-ws.org/Vol-2044/paper12

Knowledge graphs have become increasingly important in information retrieval tasks. However, if semantically interlinked concepts do not reflect the semantics of a document corpus, users might be confronted with non-relevant query results. In this work, we propose a network-metrics based method that allows assessment of knowledge graph quality within the context of a domain-specific document corpus. Preliminary results show that our methodology is able to point out structural and semantic issues in the knowledge graph, as well as provide information on the overall semantic fit of knowledge graph and domain specific corpora.


Copyright © 2018 by the paper's authors |valid HTML5|