DFKI-LT - Towards the Generation of Semantically Enriched Multilingual Components of Ontology Labels

Thierry Declerck, Dagmar Gromann
Towards the Generation of Semantically Enriched Multilingual Components of Ontology Labels
in: Paul Buitelaar, Philipp Cimiano, David Lewis, James Pustejovsky, Felix Sasaki (eds.):
3 Proceedings of the 3rd International Workshop on the Multilingual Semantic Web (MSW3) volume 936,
CEUR Workshop Proceedings, Pages 11-23, Boston, MA, USA, CEUR, MIT, Boston, Aachen, 11/2012

 
Ontologies often contain multilingual textual information in annotation properties, such as rdfs:label and rdfs:comment. While the motivation for using such annotation properties is to provide a human readable description of abstract conceptualization of the domain, we notice that the importance of appropriate natural language use and representation is often neglected. The same can be observed with resources on the Web, such as multilingual taxonomies. Terms often lack consistency and completeness, hampering also an accurate automated natural language processing of such text. We propose a pattern-based transformation of terms in labels, thereby also supporting a multilingual alignment of (sub)components of labels. The source data for our approach is an ontology we derived from an industry classification taxonomy, which we improve as regards consistency and completeness and apply to the process of lexicalization
 
Files: BibTeX, paper2-MSW3.pdf