Towards Ontology Engineering Based on Linguistic Analysis

Paul Buitelaar (1), Daniel Olejnik (1), Mihaela Hutanu (2), Alexander Schutz (2), Thierry Declerck (2), Michael Sintek (3)

(1) DFKI GmbH, Language Technology, Stuhlsatzenhausweg 3, 66123 Saarbruecken, Germany, {paulb,olejnik}@dfki.de; (2) Universität des Saarlandes, Computerlinguistik, 66041 Saarbruecken, Germany, {hutanu, schutz, declerck}@coli.uni-sb.de; (3) DFKI GmbH, Knowledge Management, Erwin-Schrödinger-Straße, 67608 Kaiserslautern, Germany, sintek@dfki.de

O1-W

In this paper we describe OntoLT, a plug-in for the widely used Protégé ontology development tool that supports the interactive extraction and/or extension of ontologies from text. The OntoLT approach aims at providingan environment for the integration of linguistic analysis in ontology development. OntoLT enables the definition of mapping rules with which concepts and attributes can be extracted automatically from linguistically annotated text collections. Mapping rules are defined by use of a constraint language. Constraints are implemented as XPATH expressions over the XML-based linguistic annotation. If all constraints are satisfied, the mapping rule activates one or more operators that describe in which way the ontology should be extended if a candidate is found.

ontologies, text mining, XML annotation

550.pdf