Jim, (01)
I strongly agree with the need for curators. (02)
JR> For an enterprise vocabulary, consistency and validity are
> important. Such vocabularies are "curated" by a group within
> the enterprise. This group has typically been responsible
> for collecting the information from subject matter experts
> (SMEs), testing it for consistency and validity, and then
> publishing it for others in the enterprise to use. (03)
But I'd also like to emphasize that the curators need automated
and semi-automated tools to assist them. (04)
JR> With the Semantic Web, there is the possibility of a different,
> less labor intensive model of curating... Semantic Web technologies
> like OWL-DL allow this information to be classified by the curators. (05)
I agree. But I keep emphasizing that the current tools are still
in their infancy. Systems like Cyc, for example, have vastly more
sophisticated tools. (06)
For years, Cyc had a reputation for being horribly slow. But the
primary reason was that their huge knowledge base (over 2 million
axioms) wasn't sufficiently modularized. They did group the axioms
into microtheories, but it was possible for stray axioms to get
sucked into the mix and cause the inference engine(s) to get
sidetracked on irrelevant tangents. (07)
But more recently, they've been doing more work on modularizing
the knowledge base and speeding up the system. On the kinds of
inferences that DLs do, Cyc is just as fast -- but it also supports
a much wider range of methods for a much wider range of problems. (08)
Even with the RDFS and OWL languages, much more can be done. For
testing consistency, many people are using Formal Concept Analysis
(FCA) to verify the completeness and consistency of the OWL and
RDFS ontologies. For more info, type the following to Google: (09)
OWL FCA concept (010)
This produces 679,000 hits, of which the top 50 are all relevant.
(The word 'concept' helps avoid extraneous hits about Nite OWLs,
Facebook, and the Fellowship of Christian Athletes.) (011)
Many people have tried to avoid multiple inheritance because it
might create inconsistencies if people use it incorrectly. But
that is why automated tools such as FCA are important: (012)
1. Given a set of concepts and defining attributes, FCA tools
generate the minimal lattice that shows all inheritance
paths. (013)
2. Every one of those paths is guaranteed to be consistent with
the given set of definitions. (014)
3. The results can be used to check whether the people who typed
in the OWL or RDFS had missed any important paths and ensure
that the ones they specified were correct. (015)
The original FCA tools (from the mid 1990s) were slow, but the
newer implementations are much faster, and they can be used to
verify large OWL knowledge bases. (016)
John (017)
_________________________________________________________________
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 (018)
|