A Cross-Lingual Correcting and Completive Method for Multilingual Ontology Labels

Dagmar Gromann, Paul Buitelaar, Thierry Declerck

In: Paul Buitelaar, Philipp Cimiano. Towards the Multilingual Semantic Web. Seiten 227-242 ISBN 978-3-662-43584-7 Springer Heidelberg, New York, Dordrecht, London 9/2014.


Multilingual content in ontologies has one of the highest potentials for bridging linguistic borders on the Semantic Web. Human readability and automated linguistic processing of Multilingual Semantic Web resources depend on natural language content represented in labels. As there are currently no standards or best practices for labeling ontologies, existing labels are frequently highly condensed up to the point of losing their domain-specific expressivity. For instance, ellipses often used in labels pose a challenge to linguistic processing. Elided domain-specific elements challenge human users and machines alike. Thus, the proposed method expands condensed labels in four main processing steps by resolving complex natural language phenomena. It heavily relies on a cross-lingual comparison and employs idiosyncratic benefits of one language to process other languages.


Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence