SOOKAT and Ontologies

The tool SOOKAT can be described also in terms of ontologies, described by Nicola Guarino [4]. Models of SOOKAT are discovered to be different kinds of fine-grained ontologies formed during the KA process:

A domain ontology

is defined by Guarino to "describe the vocabulary related to a generic domain". The DM is thus actually a subset of a domain ontology, that is needed in the KB being constructed.

Application ontologies

are defined to "describe concepts depending both on a particular domain and task, which are often specializations of both related ontologies". DGs consist of task-dependent inferential dependencies between domain concept attributes, that have been defined in the domain ontology. Although the task ontology has not necessarily been formed, the DG seems to fulfil the requirements for an application ontology.

A task ontology

is defined to "describe a generic task or activity". The IS describes the task of the KB. It is formed using a bottom-up strategy from the DG, i.e. the application ontology. The reason for this order, which reverses the one presented by Guarino, is that details of the task are defined only during the KA process.

 

 

In general, shared and stabilized domain knowledge forms the domain ontology, whereas individual know-how forms the application ontology.

In collecting value suggestions from different knowledge sources for attributes of subclasses of abstract DM concepts, the attributes can be considered predicate symbols, and the values suggested considered constant symbols of the vocabulary, together forming intentional interpretations.

Modelling and inference can in SOOKAT use a shared task ontology.

Sat Oct 14 22:52:14 EEST 2000