A pet store owner could select to use a biological
taxonomy ontology and a commercial one, and add a descriptor
field to animal called product shipping
type.
He can also create new subclasses and publish this new
ontology.
8 Ontologies, Agents, and Services
Web services are the backbone of the SW, as well as the .NET and ONE initiatives.
The SW will create machine-readable ontologies that describe services.
These descriptions will be used by "capable" agents. They
will be able to find matches up and down the class
hierarchy and follow other type of semantic links.
A machine-readable description of the service and logic
describing the consequences of its use can also be
included.
We could also use DAML+OIL to define an ontology of the terms
needed to describe the invocation of services. This could be
done by describing them as FSMs.
9 Service Class
A potential service class and its properties.
DAML-S seems to be the
only viable standard at this momemt.
10 Agents and Ontologies
Big boxes are ontologies, small boxes are agents.
Agents can use the same or related ontologies. This would allow for some degree of functionality.
11 Logic Layer
Universal language for monotonic logic.
Any rule system can export, generally cannot import.
No one standard engine - inference capabilities differ.
Many engines exist (SQL to KIF, Cycl, Jess, etc).
Any system can validate proofs.
12 Semantic Web Bus
13 Deployment
RDF extraction and report generation using XSLT.
RDF and Topicmaps convergence.
Many general purpose RDF engines.
Generic database RDF tools.
Generic and specific GUIs.
OWL ontologies and inference tools needed.
A killer app?
14 Knowledge Evolution
The SW can assist the evolution of human knowledge as a
whole.
Coordinating actions across a large group is painfully slow
and takes an enormous amount of communication.
The SW names every concept by an URI. It lets anyone express
new concepts that they invent with minimal effort.
This will make human knowledge and its workings available to
agents.