Lists Home |
Date Index |
"Bullard, Claude L (Len)" wrote:
> Describe the discovery behavior.
> a) For what reason is it initiated
Using a classic example: A travel agent Web Service needs to discover
one or more hotel Web Services - i.e. a Web Service that would be
offered by a hotel/hotel chain that has functionality required for
automatic (at least partial) reservation of a hotel room, and all of the
functions that are required by that (ability to search by specific
> b) By what mechanism is it accomplished
One way, of multiple: The travel agent Web Service discovers (how it
discovers it will be orthogonal for this example) an e-business registry
- such as UDDI or ebXML Registry - and performs an automated search of
the registry contents. We can also assume for this example (not speaking
about any particular standard) that the search would be
"semantically-aware" - i.e. that it would involve concepts such as rich
metadata attributes in registries, classification of entities according
to ontologies, etc.
> c) What entity evolves as a result of
> acquiring the behavior?
Several entities, among them:
- A "record" (or "object" - using terms loosely) that represents a hotel
Web Service in the registry, along with its pertinent business (i.e.
organization information about the hotel/hotel chain) and technical
(e.g. supported transport protocols for Web Service) information.
- A Web Service description (e.g. WSDL document) that, if a particular
hotel is "accepted" in the search (i.e. it meets all stated criteria),
will be used to "connect" with the hotel's Web Service.
> d) How is a message type shared
Through the WSDL document.
> and how
> after it is shared is it discarded/forgotten
> as a type (pruning the theory lattice)?
Not sure what you mean here - but this should be taken care of by the
stateless nature of the exchange (assuming that it is stateless - which
will change in the future with certain emerging standards).
> The realistic answer to c) is the client.
Ok - then I retract my answer from above (I was thinking along different
> In an ontologically endowed SOA, the client
> must be able to consume a service as a result
> of using the ontology (which itself can be
> delivered by a service), but it can also
> produce an ontology and return it as a
> theory to the emitter.
Yes - perhaps the produced ontology is a result of the aggregation of
multiple ontologies that each represents a specific sub-domain (e.g.
specific "types" of hotels), that is then returned to the requester for
them to - perhaps - use the aggregated ontology for the current
(in-process) search, or for later searching.
> So discovery
> is a feedback-enabled behavior such that
> the act of using it improves it and
> evolves the system.
Yes - Roger, are you there? :)
Booz | Allen | Hamilton
Strategy and Technology Consultants to the World
> A model/theory/ontology
> submitted can then be shared and such
> sharing is the semantic equivalent of
> the Google PageRank.
> SOAs are described as a framework of
> shared messages and it is the evolution
> of the message set that should be examined
> because this is the standardization behavior.
> Multiple SOAs that are ontologically distinct
> but share overlapping message types are
> information ecosystems. Such systems are
> not boundaryless. In fact, the standardization
> behavior creates boundaries and these affect
> the survival and propagation of any member
> of a distinct information community.
> Next, we must better understand the evolution
> of distinct situations which involve members
> of multiple information communities and how
> these affect the standardization behaviors
> over these messages.
> -----Original Message-----
> From: Chiusano Joseph [mailto:email@example.com]
> "Bullard, Claude L (Len)" wrote:
> > A simple question, but one that could get to
> > the heart of whether or not we should expect
> > to see growth in the semantic web business:
> > What are some services that SemWeb
> > applications could expose for discovery
> > and integration into web client applications
> > including mobile applications? In other
> > words, how and where do the Sem Web and
> > Service Oriented Architectures meet?
> This should be a good thread. I could go on and on about this, but at
> the most rudimentary level (being very simple with the language, and
> making some basic assumptions):
> - A primary foundation of SOA is shared services
> - Semantic Web services will be described in a semantically rich-enough
> manner so as to be efficiently discovered (by "efficiently", I mean both
> easily and according to the service discoverer's criteria such as QoS
> - SOA + Semantic Web = More easily discovered services that are better
> tailored to a service discoverer's (human or machine) needs, thereby
> enabling service providers to broaden their reach, and service
> discoverers to have their processing needs more easily met (higher
> "service satisfaction") - which translates into better service for their
> customers (if a customer-oriented scenario).
Booz | Allen | Hamilton