Doug,
You made my day!
My first draft of this model was published as Data Model Patterns:
Conventions of Thought in 1995. It was derived entirely from my
experience on projects, trying to model the things I encountered. When
Cyc came along, I was intrigued and impressed by their ambitions, but I
found the documentation (that I could get for free) somewhat
intimidating. The whole point of the way I do models is that they
must be understood by non-technical business people who have no prior
experience with modeling (or Greek philosophy). So clarity is kind
of important.
From what you said, "GeographicalThing and
SurfaceRegion-Tangible" does sound suspiciously like my
Geographic Location. Mine is a lot easier to present to
business people.
In my last pass at this, published in 2011 as Enterprise Model
Patterns: Describing the World, I was being a bit more ambitious, now
that I knew there was something called an "upper level
ontology". But I didn't look to Cyc. I looked to logic
and the projects I've had to deal with over the last twenty years.
>From the way you describe Cyc, I am rather pleased with myself. We
seem to have come to much the same conclusions.
In the example, I did make the assumption that a boundary was an
abstraction. The boundaries of a piece of real estate are lines,
positioned by the geographic points. These points may be defined in
terms of any of several geographic schemes. The most common are
what I call simply "geographic" (latitude, longitude,
elevation) and Cartesian (x, y, and reference point (which itself has to
be located somewhere)).
Any earth, oil, or other physical materials are in the category of
Physical Asset. Yes, these resources are located in terms of
the geography we are discussing.
And doing the geography of mountains sounds lot like Geographic
Solids.
By the way, in the later book, I actually address the problem of
describing the world in multiple levels of abstraction.
Level 1 is the basic generic model of the world, much derived from my
first book. Here, my basic categories are:
- People
and Organizations
-
Geographic Locations
- Physical
Assets
-
Activities and Events
-
Time
As you can imagine, the first four of these are very similar, so the
Level 0 of abstraction is a template for those: Thing, Thing
Type, Thing Relationship, etc.
Modeling time, as is I'm sure the case in Cyc, a special
situation.
Also at level 0, because I didn't know what else to do with them, are two
"meta" data models: Accounting, since that
is itself a model of the business, albeit in different terms than a data
model; and Information Resources (e.g. Documents,
Books, etc.) , which in one sense (at least use to
be) about physical resources, but each of these is significant because it
is about something else in the business model.
Level 2 gets a bit more concrete, assembling the Level 1 elements into
models of business functions, like contracts, facilities management (the
facilities/addresses/sites I talked about before), marketing and
communications, etc.
Level 3 consists of examples of real business situations that cannot be
addressed by the more generic structures. This includes
microbiology, banking and finance, highway design, etc. In each
case, I assume that most of the requirements of an enterprise can be
covered by the level 0-2 models. But there is something that calls
for special attention.
Doug, I am gratified to read that apparently my level 0 and 1 models
aren't terribly different from what the Cyc people came up with.
Yes, I did it myself, unconstrained by history, but I was more fun that
way.
Thanks for your comments.
Dave
At 12:34 PM 9/26/2012, you wrote:
On Wed, September 26, 2012
11:14, David C. Hay wrote:
> In my model of the world (dare I say "upper level
ontology"),
> Enterprise Model Patterns: Describing the World, I have put a lot
of
> thought into the distinctions you describe here.
When was this model created? Cyc had all of these concepts by
1998,
if not by 1996 -- other than the "Other X" concepts. (IP
address was
created in 1999). Cyc made the distinction of treating
geopolitical
entities from geographical regions in 1998. One could use a
context
in which the geopolitical entity was an agent, or one in which it
was
also a piece of land. This ontology fragment has it as a piece of
land.
The #$comment for #$GeographicalRegion reads:
'A specialization of GeographicalThing and SurfaceRegion-Tangible.
Each
instance of GeographicalRegion is a tangible spatial region that
includes
some piece of the surface of a planet (usually PlanetEarth), and may
be
represented on a map of the planet. This includes purely
topographical
regions like mountains and underwater spaces, places defined by
demographics (e.g. language areas) and territory otherwise
demarcated
(e.g. TimeZones). In "dualist" geopolitical contexts (see
DualistGeopoliticalMt), instances of GeopoliticalEntity are also
considered to be instances of GeographicalRegion.
In all cases the region in question must contain some tangible
component
with which it is possible to make physical contact. The instances of
GeographicalRegion contrast in this respect with the instances of
GeographicalThing-Intangible, which are wholly intangible.
Examples of GeographicalRegion include RockyMountainStates-USRegion,
the
ContinentOfAustralia, SinaiPeninsula, and -- in "dualist"
geopolitical
contexts -- YaleUniversity and CityOfPittsburghPA. Some important types
of
regions are represented by the sub-collections LanguageArea,
TimeZone,
PostalCodeRegion, EcologicalRegion, ConstructionSite, and -- in
"dualist"
geopolitical contexts -- GeopoliticalEntity. No instances of
GeographicalRegion are wholly indoor locations.'
I urge people who want to develop ontologies of common things to
first check what OpenCyc has. It can save a lot of reinventing
of
complex wheels. Cyc is certainly far from perfect and has huge
holes, but with hundreds of staff years being put into a publicly
available ontology, many parts of which are very carefully
considered,
it can be used to speed up such ontology development projects.
One can save staff months or years by examining what has already
been done in your field.
-- doug foxvog
> In my model, Geographic Location, is an place on Earth (OK, I am
a
> geocentric personality, but the model can be extended skyward
if
> necessary). The sub-types are Geographic Area, Geographic
Point,
> Geographic Line, and Geographic Solid. (I only encountered
the
> latter two late in my career, when I worked for a telephone
company
> and an oil company.)
>
> Geographic Area, then, is defined as any bounded 2-dimensional
place.
> (Since we're talking about the Earth, we have to recognize that
it's
> really a spherically 2-dimensional place, but that's in the
advanced
> class...)
>
> Geographic Area has four sub-types:
>
> 1. Geopolitical Area is a Geographic Area whose boundaries
are
> defined by law or treaty. Thus you have the Geographic Areas
that
> are "California", "Los Angeles County", and the
like. Defining
> sub-types for Geopolitical Area depends on where you are, but
I've
> come up with at least these:
> a. City
> b. Country
> c. Principle Country Subdivision (which is a
really ugly name for
> "State", or "Province" or (for our UK buddies)
"County") In most
> real models it turns out to be something like State or
Province.
> d. Other Geopolitical Area (subject to
negotiation).
>
> 2. Administrative Area is a Geographic Area whose boundaries
are
> defined, typically, by an organization, such as a corporation,
but
> also may be by a government. Examples of this include
"South-central
> sales area" (defined by a company), "United States Pacific
Command
> (USPACOM)"(defined by the US Department of Defense), and more
common,
> a "Postal Area" (defined by a national postal
authority).
>
> 3. Natural Area is a Geographic Area whose boundaries are
defined by
> natural phenomena, such as lakes or continents.
>
> 4. Other Surveyed Area is any Geographic Area (other than 1-3,
above)
> that is measured and whose boundaries are recorded. This
includes
> the lot my house is on, the area that is the location of the
World
> Trade Center in New York, and so forth.
>
> Geographic Point is just that: a Geographic Location that is
a
> single point. Usually it's attributes are "longitude",
"latitude",
> and "elevation", but with different systems of geographic
reckoning,
> they could be something else.
>
> Note the this model requires a Geographic Location
Relationship,
> which is the recursive entity type that links one Geographic
Location
> to another. Sub-types of this include Geographic Definition
(linking
> points to areas, lines and solids), Geographic Structure (that
puts
> "Boston" inside "Massachusetts"), Geographic
Overlap (one of which
> puts the "Navaho Indian Reservation" in
"Arizona", and another of
> which puts the "Navaho Indian Reservation" in New
Mexico. And of
> course there's Other Geographic Location Relationship, just in
case.
>
> All of this is a pretty sophisticated way to represent just
> places. The Government that has jurisdiction over a
Geopolitical
> Area is represented by an Organization, linked to the
Geopolitical
> Area via a link class that I have cleverly called Jurisdiction.
>
> Note also that this only covers the place itself. A
"place with a
> purpose" is variously called a Site, a Facility, or an
Address. This
> is where one or more Parties (Persons or Organizations) are
located
> to perform one or more Activities, which consume, use, or produce
one
> or more Physical Assets.
>
> Note that Address (for example) has two sub-types: Physical
Address
> (which has some fairly complex relationships with Geographic
> Location), and Virtual Address (which nicely covers Telephone
> Address, E-mail Address, IP Address, and so forth.)
>
> A Party may be located in one or more Addresses (Physical
or
> Virtual), just as an Address may be the location of one or more
Parties.
>
> Note that a physical building is located in one (or more?) Site
> (again, for example), but it is not the same as the Site. It
is a
> Physical Asset.
>
> Among the simplest applications we build simply lists names and
> addresses. That's an output. To actually understand what
goes
> inside that "simple" application requires way more
sophistication
> than most people appreciate. I've spent a fair amount of my
career
> working on this model.
>
> If you've actually read this far, you are a good candidate for
buying my
> book.
>
> Regards
>
> Dave Hay
>
>
> At 08:52 AM 9/26/2012, you wrote:
>>On 9/26/2012 8:53 AM, Andries van Renssen wrote:
>> > The reason why the _expression_ 'I dig a hole in the school
district'
>> > sounds odd is: because that _expression_ is a short-cut for
'I dig
>> > a hole in the land that has a role as school
district'.
>>
>>I agree with that point. The notion of role is essential
for
>>distinguishing every subdivision on planet earth. There
is
>>always a reason or a purpose for the choice. That is true
for
>>everything from countries and continents to things like
farms,
>>parking lots, and playgrounds.
>>
>> > But the piece of land that is defined by that boundary
is
>> > nevertheless a physical object, and it has a mass,
although
>> > its value is unknown and not of interest.
>>
>>Space is physical, but it doesn't have a mass. An area
is
>>a two-dimensional region. The political subdivisions
only
>>specify coordinates that determine the area at the surface,
>>and they are silent about depth or height.
>>
>>By fiat, the governments of countries lay claim to the
mineral
>>rights beneath their areas. In principle, they could claim
rights
>>down to the center of the earth. But in practice, the
technology
>>can only mine a few km. beneath the surface.
>>
>>When air travel became possible, national governments laid
claim
>>to the air space above them, but smaller governments did
not.
>>But nobody laid claim to the regions above the atmosphere.
>>Those are more distinctions by fiat.
>>
>>In summary, I recommend that any ontology for any
subdivision
>>of the earth should specify the surface area S and the
intended
>>role R for that area.
>>
>>Then anything else that may be associated with the pair
(S,R),
>>such as the land, air, water, people, buildings,
governments,
>>should be specified as the X associated with the area S as
>>considered in the role R.
>>
>>John
>>
>>_________________________________________________________________
>>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
>>
>
>
_________________________________________________________________
> 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
>
_________________________________________________________________
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
_________________________________________________________________
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 (01)
|