On Fri, April 6, 2012 16:08, John F Sowa wrote:
> On 4/6/2012 10:27 AM, doug foxvog wrote:
>> I think the issue should be addressed. Instead of simply deleting
>> the passage, it should be replaced. (01)
> I agree that the issue is worth discussing. But it's so complex that
> it's hard to summarize in short bullet items. (02)
Agreed. (03)
>> 3. Ontologies can be built to have both expressivity and scale.
>
> There are some important distinctions hidden beneath that statement.
> An ontology is a theory about what exists, but not every theory
> is an ontology. And a give ontology might consist of a huge number
> of interrelated theories, microtheories, alternatives, and special
> cases. (04)
Agreed. (05)
> But even a very complex ontology might be used as a resource from
> which many special cases can be derived. ...
> Then there are the questions about who or what is going to do
> those transformations and adaptations. The SME? The knowledge
> engineer? (06)
Who does the programming -- the programmer. Who does the
extract from a complex ontology -- the knowledge engineer. (07)
> The user who has the problem? A consultant hired
> for that purpose? (08)
A KE could be a consultant. (09)
> Some tool(s) that are used by or triggered
> by any of those people? Anybody else? Why? (010)
Once such a tool gets designed, built & tested, it could certainly be
useful. (011)
>> Although an ontology may be highly expressive, uses of that ontology
>> need not calculate every statement that could be derived from the
>> rules in the ontology and the data in the local knowledge base. (012)
> Are you talking about the ontology, a theory derived from it, or
> an implemented program or system that is derived from a theory that
> is derived from the ontology? (013)
I'm referring to a system that uses an ontology in encoding a knowledge
base. Such a system could use a more limited logic than the ontology
and not derive information that could be derived from the deeper logic.
Even if it had the logic to perform the reasoning, it could treat various
rules as backward chaining (i.e., they'll fire only when the system is
trying to prove their consequent.), and not execute the rules. (014)
>> Just because you CAN calculate something doesn’t mean you SHOULD
>> calculate it. Stick to information that is strictly useful to building
>> your big data application. (015)
> End users don't know anything about what is being calculated or why. (016)
The ontological engineer who hooks up to the ontology should design
the connection such that only the conclusions that the SMEs want
to be made are made. (017)
The end users should not make this decision. They should be
able to ask questions of the system, though. (018)
> SMEs are experts in their subject, not in any kind of calculation. (019)
They should be expert in the sort of calculation performed in their
field of expertise. (020)
> Who or what is making those calculations? Why? (021)
The inference engine would make those calculations according to
the way the system is programmed in order to solve the problem
which the users and SMEs deem important. (022)
John, these statements are made to contradict the claim that
a complex ontology necessarily would result in excessive computation.
It depends upon whether the systems using the ontology decide to
perform such complex calculations. (023)
>> Computationally expensive rules need not forward chain. (024)
> An if-then statement has no implicit computation associated with it.
> When it is used as a rule, the same statement can be used in many
> different ways for different purposes to solve many different kinds
> of problems. The qualifier "computationally expensive" cannot be
> applied to any if-the statement taken out of context. (025)
Agreed. (026)
>> Backward chaining complex rules need not be computed in a knowledge
>> repository. Such calculations can be performed locally to answer
>> questions and the results can be stored back into the repository.
>
> Again, any talk about backward chaining is getting down to nitty
> gritty computation that is many levels removed from the ontology.
> Who are you talking about? What is their purpose? (027)
I am addressing the concern that an ontology with rules would
necessarily slow things down. The OEs can design the systems so
that such computations are done only when desired -- just as
programmers do with code! (028)
>> There can be a lot of value just getting the data federated and
>> semantically aligned. (38GD)
>
> I'll give a qualified yes to that point. The major qualification
> is that I have no idea what you are federating or why. All of the
> above qualifications apply. (029)
Of course, I left this in because i agreed in the same qualified way you
did. (030)
-- doug f (031)
> John (032)
_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2012/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2012
Community Portal: http://ontolog.cim3.net/wiki/ (033)
|