In this thread, I mostly agree with the remarks by Ed Barkmeyer.
I might find a few to quibble with, but I'd like to start by
highlighting one point: (01)
RC>> Past history indicates that the killer app comes first, and
>> the standards follow. (02)
EB> I think that was John Sowa's original point. (03)
Absolutely! And I keep emphasizing the point that Cyc has been
searching for that killer app since 1984, and they haven't found it.
Millions of research dollars have been poured into Cyc, they hired
some of the best people in AI, knowledge engineering, linguistics,
logic, comp. sci., and related fields. They pioneered the term
'ontology engineering', they have produced the largest formal
ontology on planet earth, and they had some top corporations and
gov't agencies work with their ontologies. (04)
But after 24 years, nobody has found an application that can
produce enough ROI to keep Cyc in business without a continuous
stream of research grants. In the early 1990s, Microsoft was one
of the sponsors that gave Cyc enough money to license all their
software and ontology (of that time) for unlimited use in any or
all MSFT products. But MSFT terminated the license because they
couldn't find any useful application for Cyc. (05)
I firmly believe that ontologies are useful and that continued
research is important. But the evidence of Cyc indicates that it
is premature to legislate a standard ontology. Instead, we should
start with a library, repository, or registry of ontologies that
various groups contribute. We've discussed this before, but the
organization, management, and policies of the library are topics
for another thread (or threads). (06)
PC> I can take two ontologies both based on FOL, OpenCyc and SUMO,
> and I cannot envision any automatic method to accurately determine
> whether one element in OpenCyc is the same as, subtype, supertype,
> disjoint, or overlapping with a SUMO element, because the basic
> vocabularies are different and need human inspection to determine
> the intended meanings of the terms. Even then, it may be
> impossible because the representations are not detailed and the
> documentation can be inadequate to resolve ambiguities. (07)
That is true. That is why I have emphasized the need for the people
who contribute, use, or review the ontologies in the library to say
what relationships they find among the ontologies. I have often
used the term 'lattice of theories', but I'll weaken it to just
'partial order', since I don't expect a complete lattice to be
implemented. As an informal term, I'd say 'hierarchy of ontologies'. (08)
AT> Long story short: I really believe that this [meaning primitives]
> is an interesting question and needs to be evaluated with rigor.
> We may find that there is no general solution, but many practically
> applicable special solutions, as pointed out by Pat. (09)
I agree. But it's an idea that has been promoted in computational
linguistics since the 1950s and in philosophy for much longer --
and there still isn't any consensus. It's an important research
topic, but not one that is sufficiently stable for a standard. (010)
SB> I think the problem of creating standards is not about being
> pro-active or reactive, or about testing what has been produced,
> or starting with existing technologies. It is about two way
> communication between modellers and domain experts. (011)
I agree. I was criticizing the proactive approach because some
people in this forum seem to be too eager to standardize something
proactively when there does not appear to be any consensus among
experts, modellers, and users. (012)
RC>> While I can imagine *formally* what it would mean to have such
>> an ontology -- a single theory of time/space/matter/motion/
>> properties/<YourFavoriteFundamentalConceptHere> of which all
>> other ontologies are extensions -- I think Ed and others have
>> raised reasonable doubts about whether it is a realistic prospect. (013)
PC> OK, there are those who are not yet convinced it is feasible... (014)
The main reason why people aren't convinced is that there never
has been any single *best* theory in any subject. Every branch
of science and engineering -- including physics -- has an open-
ended collection of approximations, each specialized for various
problems and purposes. It is, however, useful to organize those
theories in a hierarchy -- and that's what I believe we should do. (015)
AA> Thus to build an ontology standard, we need start constructing it
> by formulating and selecting the most viable ideas.... (016)
I agree. Put them all in the hierarchy, show how they're related to
one another, and let the users discover which are the most useful
for various purposes. (017)
AA> But what is essential now to decide on the viability of the concept
> of Federated Ontology System, which, to my experience, looks more
> promising than any nonfederated ontology systems, either unitary or
> centralized or loose and unconnected, which sorts currently prevail
> and propagate as pandemic on the WWW. (018)
If the federated ontologies are organized in a hierarchy, I would
consider that a reasonable way to start. (019)
NC> In my opinion if you are waiting for "... strong leadership by
> managing organizations .... in particular government organizations
> when it comes to spending tax dollars to the maximal benefits of
> the people, and not just one project", then you are a bigger
> optimist than anyone on the forum. (020)
I spent 30 years working for IBM, which was considered one of the
best managed companies on the planet at the time. They had many
excellent researchers who were at the forefront of almost every
major innovation. But somehow, the management would consistently
dissipate the advantage. Just look at SQL -- developed at IBM,
but commercialized by Oracle. Another example is the PC. Every
researcher at IBM groaned when the idiots at Boca Raton chose the
Intel X86 over the Motorola 68000. Intel had their "lobbyists"
at IBM who sweetened the deal to cut out Motorola. (021)
That is one reason why I put more trust in the users than in any
kind of centralized authority. It's much easier for the lobbyists
to convince (i.e., buy) one "decider" than a million users. (022)
RW> I think that there might be justification for certain industry
> groups to get together to produce their own "standard" ontologies. (023)
I agree. But one problem is that they are all highly protective
of their "trade secrets", and they don't want to give up anything
that might be a competitive advantage. Sometimes an industry group
might coalesce in a consortium or a research institute on "neutral"
ground, such as a university. That should be encouraged. (024)
FMcC> The compiler is still in its early stages, with very little
> code generation and very little in the way of optimizations
> implemented. But already, there is a 7-2 ratio in post parsing
> to parsing code. (025)
That ratio is consistent with my experience, and it illustrates
what I believe is one of the worst policies adopted for the SemWeb:
focus on a one syntax-fits-all notation in order to promote reuse
of a horribly limited and inefficient parser. (026)
CR> In order to better define, test, and evolve our standards, the
> OGC uses a test bed process. This rapid engineering design, build,
> test process definitely improves the quality of a candidate standard
> before it comes into the OGC standards process, where a formal
> process involving committees takes place. (027)
That is useful. But I'd also like to see some practical applications
that have a positive ROI before anointing something as a standard. (028)
John Sowa (029)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (030)
|