[SCL] Re: ontologies

[Murray Altheim]

Jon Awbrey wrote:
> o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o
> 
> murray,
> 
> could you explain this more simply to me?
> i begin to detect in the surroundings an
> old mare's nest about "asserted" versus
> merely "contemplated" sentences, but
> i can't quite see where the knot
> is being tied this time.

Okay then. I'll try to figure out what you meant.

> Murray Altheim wrote:
[...]
>>I have a set of documents that are expressed in Linear Topic Map
>>(LTM) notation. They gets parsed into my system and can be used for
>>their intended purposes in organizing information. Any of them can
>>be reserialized into XML Topic Map (XTM) notation, and it's certainly
>>possible to write a serializer to create an OWL or KIF or CGIF
>>document from them, or whatever. All this suits my purposes just fine;
>>I can do various sorts of inferencing (such as across transitive
>>relationships, slot-based reasoning, and I'm just beginning to work
>>on datatyping, measurement units and mereological relations).

That much I believe is fairly straightforward. Just your typical,
bread-and-butter conversions, serializations, etc. from one format
to another, using the syntax and conventions described by the
system. Jack Park has talked about hooking up Jena2 to Ceryle so
he could play in the RDF world, etc. This kind of thing has been
going on for years. It was KIF a decade or more ago, nowadays it's
RDF, OWL, Topic Maps, etc. Same pancake, different syrup. Or something
like that...

>>To anyone in the W3C or here at KMi, either the individual documents or
>>the set of them taken as a whole would be called an "ontology". As a
>>whole, it'd be a fairly big document, depending on serialization, but
>>of the order of many megabytes.

This?  I'm just stating that the sum of statements about a domain
being made are called an "ontology".

>>Now, those individual documents, or even the large document as a
>>whole, could be considered a "set of sentences". They define terms
>>that I use (as Topics in the Topic Map sense) and relations between
>>them (as Associations in the Topic Map sense). I have the ability to
>>express context at a number of levels. A fragment from one of these
>>documents is shown below (in LTM notation):
>>
>>   /* things that exist in space ....................................  */
>>
>>   [SpatialThing ; Thing = "Spatial Thing"
>>     @"http://purl.org/ceryle/authoring/#SpatialThing"]
>>   {SpatialThing, Description, "Things that exist in space, having a
>>   physical presence and location (even if in constant motion)."}
>>
>>   /* things that exist in time .....................................  */
>>
>>   [TemporalThing ; Thing = "Temporal Thing"
>>     @"http://purl.org/ceryle/authoring/#TemporalThing"]
>>   {TemporalThing, Description, "Things that exist in time, such as events or actions."}
>>
>>   [Object ; SpatialThing ; TemporalThing = "Object"
>>     @"http://purl.org/ceryle/authoring/#Object"]
>>   {Object, Description, "A thing manifesting a physical existence
>>   in the world, i.e., a physical object such as a person, a rock,
>>   a grain of sand, or even the air. Objects exist in both time and
>>   space. The physical existence of Things like spirits or demons is
>>   not addressed by this ontology."}

Point here was that while the "ontology" might be 16MB, it's
constructed of individual statements about terms and relations,
as above, just as a novel may be 800 pages but it's still
composed of sentences.

>>Now, to use the parlance now prevalent here lately, I don't give a
>>rat's ass about the correctness of the above content. But would it
>>be correct to consider them (and by extension, the larger document
>>they are extracted from) as a "set of sentences"?

Just asking the question I assumed to answer in my previous comment.

>>Would an individual
>>"statement" (e.g., the one being made about SpatialThing as being
>>a subclass of Thing, having a name "Spatial Thing", a canonical
>>identifier of "http://purl.org/ceryle/authoring/#SpatialThing", and
>>the given description) be considered as a "theory" or an "ontology"?

So, is the statements you see above in LTM syntax about "SpatialThing",
"TemporalThing" and "Object", are those statements (let's pretend there's
about two dozen individual assertions there, because that's about how
many there are) considered a "theory" or an "ontology"? Would the
statements about solely "TemporalThing" be a "theory" or an "ontology"?
I'd say that there are about three sentences, about 20-24 assertions
made by those three sentences, that *if encapsulated as a grouping*,
that would be a "theory", and taken as a whole, an "ontology". And
the definition of "whole" would be modular, since modularity is not
semantic but structural, i.e., you could have my authoring ontology
in one document, seven documents, or seventy, and they'd when processed
become the same thing. The question (to me) is whether you could
consider any "set of sentences" an ontology. I think of a "theory" as
something smaller than that, maybe some form of specific hypothesis
about something in a domain, e.g., an assertion about the relationship
between two terms.

>>I'm happy just calling it a statement or a sentence, which is what
>>it's to me closest to grammatically. I'm also happy considering that
>>the information in my document about "Spatial Thing" might be thought
>>of as my "theory" of the concept. I'm not happy with the idea that
>>this small subset of my document might be called an "ontology", since
>>I think of that as a larger thing.

I may sound like I'm contradicting myself, but I'm trying to state
that "ontology" needs to be big enough to be assertions about a
domain of terms and relations rather than a single term. The latter
to me sounds more like a theory.

>>Now, I've already broken my entire authoring ontology (which currently
>>consists of about a dozen smaller documents or sub-ontologies) into
>>"modules". The entire package I think of as an ontology, and I think
>>of the modules as *perhaps* ontologies in their own right, e.g., the
>>one on characters (i.e., fictional characters) is what I "know" about
>>characters. They are all still "sets of sentences". I'd prefer we
>>(or since I seriously doubt my opinion matters here much, you) use
>>"theory" for small things and "ontology" for big things, with some
>>notable threshold making them distinct.

Just noting that modularity is structural, not (at least necessarily)
semantic, though it could mirror the semantics.

>>Now, if all of this just confirms me as a total idiotic standards-wonk
>>boneheads who doesn't know my A from my elbow, I'll just crawl back
>>into my total idiotic standards-wonk boneheads who doesn't know my A
>>from my elbow hole and go back to work on making something actually
>>work, regardless of its correctness in either terminology or ontology,
>>and leave you guys to continue to thrash this about for the next year
>>or so.

This probably doesn't need interpretation unless one or two of the
words I used were misunderstood. Bonehead, perhaps?

Murray

......................................................................
Murray Altheim                    http://kmi.open.ac.uk/people/murray/
Knowledge Media Institute
The Open University, Milton Keynes, Bucks, MK7 6AA, UK               .

                    US STATES OF TERROR ALERT
                         green (low)
                         blue (guarded)
                         yellow (elevated)
                         orange (high)
                         red (severe)
                         purple (pucker)