Wg/linguistics/ontology building for linguistic resources

Ontology Building for Linguistic Resources
This page needs to be further extended, currently, it is just a stub.

Background
http://wiki.okfn.org/Working_Groups/Linguistics/How_to_contribute

Why is an ontology needed for (linguistic) linked data?

An OWL Ontology specifies
 * the semantics of RDF data
 * in other words: the axiomatic properties of RDF data

See also Ontology (Information Science) on Wikipedia

Basic Information on RDFS/OWL Ontologies
An OWL document contains optional ontology headers (generally at most one) and a list of axiomatic statements. These statements express constraints about the following elements:
 * categories, e.g. Tense -> owl:Class
 * relations, e.g. hasTense -> owl:ObjectProperty and owl:DataProperty
 * individuals (or objects) -> owl:NamedIndividual

In other words, an ontology consists of any number of
 * class axioms
 * property axioms
 * facts about individuals.

Naming Conventions:
 * use uppercase identifiers for owl:Class
 * use lowercase identifiers for owl:ObjectProperty and owl:DataProperty

Reference Links on RDFS, OWL

 * OWL Web Ontology Language Overview
 * OWL Web Ontology Language Reference
 * RDF/XML Syntax Specification

RDFS/OWL Ontologies for lexical resources
Feature Structures are the fundamental data structure to represent linguistic data in general and lexical data in particular, see paper by Laurent Romary http://hal.inria.fr/hal-00762664.

This is how the semantics of feature structures can be encoded in an OWL ontology:
 * Features Structure is represented as owl:Class
 * Feature is represented as either owl:ObjectProperty and owl:Class or owl:DataProperty, depending on the type of the feature values
 * Features with values that can be enumerated are represented as owl:ObjectProperty and owl:Class.
 * owl:Class is used to reference and group the possible feature values (as its individuals) and owl:ObjectProperty is used to assign the feature values defined by owl:Class to individuals
 * Example: owl:Class Tense lists values "present, past, future", owl:ObjectProperty hasTense assigns the feature values to individuals
 * Features whose values are of a simple data type (e.g. string) are represented as owl:DataProperty
 * owl:DataProperty is used to specify a simple data type as feature value
 * Example: owl:DataProperty hasWrittenForm specifies the values of this feature to be of the simple data type string

Examples of RDFS/OWL ontologies for lexical resources in RDF

 * http://www.monnet-project.eu/lemon