On Thu, August 9, 2012 09:03, John F Sowa wrote:
> On Thu, August 9, 2012 Juan wrote: (01)
>> I have encountered some difficulties concerning the meaning
>> of these terms and how they are related each other.
>
> Those are closely related terms, and nobody has ever stated
> precise definitions that could distinguish them.
>
> The oldest of the three is 'knowledge base'. It became popular
> in the late 1970s and early '80s to distinguish AI systems,
> especially expert systems, from the more familiar databases.
> ...
>
> In philosophy, the word 'ontology' means the study of existence.
> A specific ontology is a theory about what exists. I used that
> word my 1984 book, _Conceptual Structures_, but always in the
> philosophical sense.
>
> In the late 1980s, Doug Lenat coined the term 'ontology engineering'
> as a variation of 'knowledge engineering', and he advertised for
> *ontology engineers* for the Cyc project. At that time, the term
> was mildly humorous or mildly startling. But in the 1990s, it
> became more popular.
>
> The basic idea is that an ontology goes beyond a taxonomy
> of everything that exists (or can exist) in some domain.
> The crucial addition is a *theory* about what exists.
> That theory determines the critical axioms and definitions
> that distinguish a knowledge base from a database.
>
> But that definition leaves the distinction between an
> ontology and a knowledge base very unclear. Are all the
> axioms and definitions of a knowledge base part of the
> ontology? Or only some of them? Where do you draw the
> line to distinguish them?
>
> This question has been hotly debated, and nobody has a clear
> answer. (02)
> ... another issue: where do you draw the line between
> the logic used to define the terms, and the logic used for the more
> detailed reasoning. That is a controversial issue. Anybody who
> answers it uses their favorite technology to make the distinction. (03)
At Cycorp, as an "ontological architect", i worked on developing such
distinctions. We wanted basic definitions of terms that would work with
different theories (e.g., relativistic physics, Newtonian physics, and
a naive every-day understanding of physics), the different theories,
and then information using the terms and theories. (04)
We considered an ontology to define terms, properties of the terms,
and theories about the terms, while a knowledge base was like a
database, using terms defined and provided rules in the ontology
to describe information about individuals in some domain of concern. (05)
Cyc has a complex system of contexts called "microtheories". What we
came up with was three classes of microtheories to handle the above
issues: "Vocabulary microtheories", "Theory microtheories", and
"Data microtheories". The Vocabulary microtheories had the basic
definitional statements -- not much more than the subclass hierarchy
for classes, argument restrictions and maybe subsumption hierarchy
for relations (plus result types for functions), and comments for
everything. Theory microtheories had the rules; and Data microtheories
had instances of those classses and statements using that vocabulary
applied to those instances for a specific context. Several Theory micro-
theories could use the same Vocabulary microtheory with different
rules, and large numbers of Data microtheories could use the same
Theory microtheory. Vocabulary microtheories were supposed to
inherit only from other Vocabulary microtheories; Theory micro-
theories were supposed to inherit from both Vocabulary microtheories
and Theory microtheories; and Data microtheories were able to
inherit assertions from Vocabulary, Theory, and other Data micro-
theories. (06)
A Data microtheory could have rules that were applicable for that
specific context (e.g. rules for a specific partially robotic factory
that mentioned specific tools, locations, jobs, etc.), but most rules
were expected to be in the Theory microtheories. (07)
See my discussion of this in the chapter on Cyc, in "Theory and
Applications of Ontology" (Volume II: Computer Applications). (08)
Although Cyc didn't officially define the terms "knowledge base" and
"ontology" in such a way, we generally considered the Theory micro-
theories to be ontologies, and the Data microtheories to be knowledge
bases. I.e., a "knowledge base" was similar to a database. (09)
> Finally, 'knowledge model' is a term that is related to the term
> 'data model', which developed in the database field. In DBs,
> the distinction was about the storage method: in tables for the
> relational model, networks for the network model, or trees with
> cross references for the hierarchical model. Each of those three
> models has exactly the same expressive power, since anything stated
> in one can be translated to the others. (010)
This term was not used at Cycorp while i was there. (011)
-- doug foxvog (012)
> For knowledge bases, it's not clear how to distinguish a knowledge
> model from an ontology. And since the distinction between an
> ontology and a knowledge base is unclear, it's even harder to say
> what a knowledge model could be.
>
> For more about these issues, see the slides I presented in June
> for a tutorial at the Semantic Technology Conference:
>
> http://www.jfsowa.com/talks/kdptut.pdf
> Knowledge Design Patterns
>
> John Sowa
>
> _________________________________________________________________
> 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
>
> (013)
_________________________________________________________________
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 (014)
|