Tools - Enabling Semantic Web Services: The Web Service Modeling Ontology - page 136

• Logic-based reasoning tasks: reasoning tasks that are not specific to on-

tologies but are usually performed with other kinds of knowledge bases.
As the most relevant examples of such reasoning tasks, we consider:
-

instance retrieval;

- logical entailment;

- consequence finding.

• Ontology-specific reasoning tasks: reasoning tasks that refer specifically

to ontologies and their modeling primitives. Typical examples from the
description logic community are:
-

concept satisfiablity: to check whether a concept is satisfiable;

- concept subsumption: to check whether a subsumption relationship

holds between two concepts;

- relation subsumption: to check whether two relations are subsumed by

each other;

- classification tasks: for example, to compute a hierarchy of named con-

cepts in a knowledge base (knowledge base classification), or to identify
the most specific concepts.

• Reasoning tasks specific to other top-level elements in WSMO: this in-

cludes tasks related to Web services, goals, and mediators. Examples would
be:
- goal-capability matching;

- choreography matching.

Many of the reasoning tasks mentioned in the third class are in fact re-
ducible to tasks of the first and second class. Furthermore, note that there
are many more reasoning tasks, such as abductive reasoning, that have not
been explicitly mentioned. In fact, the current framework focuses only on the
reasoning tasks associated with instant retrieval.
From an implementation perspective, various, different reasoners are al-
ready available for the tasks outline above; particulary, for the well-understood
reasoning task of instance retrieval for use with the rule-based WSML variants
(WSML-Core, WSML-Flight, and WSML-Rule). More precisely, WSML-Core
and WSML-Flight can be mapped to Datalog, i.e. from the function-free Horn
subset of first-order logic to Datalog ¬ (Datalog extended by a default nega-
tion). WSML-Rule can be mapped to Prolog (Datalog ¬ extended by function
symbols). WSML2Reasoner simply integrates different reasoners into a com-
mon framework that uses WSML as a concrete syntax. Figure 10.1 illustrates
how existing systems are integrated into the framework.
The framework defines a general API for creation and interaction with a
WSML reasoner of any kind. WSML ontologies are transformed to a syntac-
tically simpler form and finally to a pure logical representation in terms of
Datalog rules. The specific semantics of the object-oriented modeling primi-
tives in WSML is captured by auxiliary rules. An API for wrapping external
Datalog engines has been defined and instantiated for a few systems. A spe-
cific auxiliary data structure that can be reused for different wrappers is a