A Practical Approach on Creating a Restricted Ontology for Crop Wild Relatives

Resumen

The task to identify a subset of about 400 terms highly relevant to crop wild relatives was performed as a continuation to an earlier project where a set of about 11400 term was extracted from on-line sources. Terms with high relevance were grouped into themes, roughly corresponding to Agrovoc (top level) categories, or indicatives of the thematic sources from which the terms were collected (biological, geographical on-line dictionaries etc), with an attempt to balance the number of terms between the groups. For the import into the ontology structure the themes were converted to namespaces in order to preserve the grouping and allow manipulation within ontology client programs on terms based on namespace grouping. Before the import the namespaces were slightly modified and adapted to some other existing ontologies. In addition to selecting relevant terms and definitions, definition of vertical and horizontal relationships between the terms was performed. Terms were also linked to sources (uris) through Dublin Core extensions of the ontology structure. The export from sql to rdf/owl was done with a script written in Perl that extracted the terms, descriptions, sources from the database, plus vertical hierarchy, term synonyms and other variants, as well as some simple horizontal relationships and produced an import file for Protégé in rdf/owl format. The resulting subset of terms was provided as a number of files; a main file containing core structure, object and data type definitions and term data in separate files per namespace, suitable for import by an ontology client program such as Protégé.

Hulden, Morten A Practical Approach on Creating a Restricted Ontology for Crop Wild Relatives., 2007 [Report]