John F. Sowa wrote:
> Mike, Doug, and Deborah,
> MB> We are developing a formal semantic model of terms in the
> > financial services industry, and to do this we are using
> > the underlying concepts of OWL but restating these in English.
> That sounds like a good strategy. Do you have tools to do the
> mapping in both direction?
Sadly no. We are constrained to only be able to do what you can do with
a UML tool, albeit a very configurable one. As it is, I think I have
broken something since the model content won't even round trip within
the same tool. (01)
You looked at the model a long time back and suggested that some of the
relationship names have the potential to be interpreted as a kind of
controlled English, and I've taken encouragement to try and be
disciplined about the naming of those. (02)
> MB> We refer to the OWL Classes as "Things" and properties as
> > "Facts" namely "Relationship Facts" and "Simple Facts"
> > respectively. These are modeled in a UML modeling tool from
> > which we produce both diagrams and spreadsheets, for review
> > via a website (this is at www.hypercube.co.uk/edmcouncil ).
> The general principle sounds good, but "things" tend to be nouns.
> What about concepts represented by verbs and adjectives?
Well, the English language was rapidly running out of words that didn't
already have some technical meaning, for instance your earlier
suggestion of "Entity" might be semantically more accurate for the top
level concept. However, not only was "Thing" sitting there more or less
unattended (and it caused a lot of comment when I started using it among
our SMEs), but also it is the exact same word used in OWL for the class
at the top of the model taxonomic hierearchy. (03)
For the sake of clarity, when I refer to something a "Thing" in the
context of the EDM Council Semantics Repository, I mean the Universal
Quantifier, the upside down A. So it is the super-set of everything in
the model. (04)
So what about verbs and adjectives? (05)
I did take another leaf out of your book here. I know you said that the
KR Lattice partitions were only examples of a possible partitioning, but
took a more fundamentalist approach to this (and this was long before
you made that comment). As I see it, any high level partitioning should
be traceable back to some reasonbly well attested philosophy, and not
made up on the fly by some ontologist. So I used your KR Lattice
partitions as is, having followed your development of these concepts in
the philosophy. (06)
As a result, one pair of partitions is the Continuant / Occurrent
partition. Now, I have got into a lot of trouble in the past for using
those two words in their simplest sense of describing two kinds of thing
(things which have some ongoing existence and things which simply
happen), without being aware of the long history of other people's usage
of those words to conjure up all sorts of four dimensional magic and
other intellectual man-traps. For the sake of clarity then, when I say
"Continuant" I mean something that has some ongoing existence, whether
or not that something is modeled in a three dimensional or a four
dimensional modeling framework. I am trying to describe the thing itself
and not the modeling framework. (07)
So, I mention this because, having introduced a partition of occurent
things, we have the capacity to extend what is modeled in the ontology.
Looking at this from the perspective of the Zachman Framework or of
standard UML theory, there are two kinds of things which might be
modeled, namely behavioural and structural, and so far, when we consider
an ontology with "Thing" at the top, and define Thing as being only
those things you can paint white or salute, then an ontology seems like
something for modeling the structural domain but not the behavioural
domain. Adding the "Occurrent" partition however, (under the Universal
Quantifier), means that in principal we can also do behavioural models,
and so for example I have done a number of business process flow models
of securities issuance, in which each process activity is also a kind of
"Thing" which is a Process Activity, which is an Activity, which is an
Event, which is an Occurrent Thing, which is a Thing. (08)
So that's kind of as close to verbs as we get. Apologies for the
slightly didactic language by the way, but that's the only way I can
seem to get all those thoughts down in front of an email audience. (09)
As for adjectives, aren't those covered by properties? (010)
Your question does remind me of an interesting point though, which is
that in natural languages we already have the means to represent
meanings, so the closer our efforts get to natural language, the more
likely it is that we have captured everything we need. Should we be
re-thinking the ontology expression format, moving away from continuant
+ occurrent things and the facts about them, and into a
Noun+Verb+Adjective framework? I don't know if that's needed for a
simple thing like describing the universe of financial securities, but I
think if one were to get into more AI related implementations, you'd
surely need more language particles like prepositions, different verb
tenses, imperative tense and so on. But as far as describing securities
in a way that can be used to help in data integration and
implementation, messaging and so on, all we need are the semantic
equivalents of the "Terms" that are already in existence in financial
industry data models, i.e. what the information industry folks
hereabouts call "TDR" - Terms, Definitions and Relationships. (011)
> By the way, the draft OMG standard for "a Foundational Subset
> for Executable UML Models (FUML)" is defined in Common Logic:
Thanks, I'll take a look at that. (012)
> DF> There seems to be two definitions of a taxonomy used in this
> > list.
> > 1. A tree structure of subtypes
> > 2. A DAG (directed acyclic graph) of subtypes.
> > I'm more comfortable with using the term "taxonomy" to refer
> > to #1 and "hierarchy" to refer to #2.
> It's necessary to distinguish structure from content. The words
> 'tree', 'DAG', and 'hierarchy' refer to structure, independent
> of what the nodes of the structure represent.
> The terms 'taxonomy' and 'ontology' refer to the content, which
> could be represented by various kinds of structures. For example,
> a taxonomy of living things is usually a tree because the normal
> way genes are inherited creates a tree. But a taxonomy of tools
> could be a DAG if you have some kinds of multipurpose tools.
> DM> Faceted classification is a big plus of Omniclass. It allows
> > something like a brick to be the brick itself, part of a wall
> > assembly, part of a hospital, from a manufacturer in Ohio.
> > Hospitals can also be made of other kinds of wall assemblies
> > so there are no straight lines or consistent elements for
> > every project or building type.
> Any ontology that is adequate for describing physical objects
> must distinguish an open-ended number of relations:
> 1. Type/Subtype for the basic hierarchy.
> 2. Type/Instance for the physical objects of any given type.
> 3. Whole/Part for types that are constructed from other types.
> 4. Object/Material for types that are made of certain kinds
> of material.
> 5. location, point-in-time, duration, mathematics, etc.
> Facets are monadic predicates or relations, and a faceted
> system allows arbitrary conjunctions of facets. That is
> a very simple and useful representation for many purposes.
> But eventually you need more general relations.
> For example, you can bundle lots of other structure into
> a single monadic relation, such as isMadeOfWood, isFromOhio,
> isThreeStoriesHigh, wasBuiltTenYearsAgo, isUsedAsHospital,
> or isOnTheOppositeSideOfTheStreetFromHospital. But
> eventually, this gets to be borderline ridiculous.
> That is why you need to embed any special-purpose notation
> within a larger, more expressive framework.
> When description logics were invented in the 1970s, they
> were used in the T-box (terminology component) to define
> a type hierarchy. But the complete system also contained
> an A-box (assertion component) that used a more expressive
> logic, such as a rule-based (Horn clause) language, full
> first-order logic, or more.
> 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
89 Worship Street
London EC2A 2BF
Tel: +44 (0) 20 7917 9522
Mob: +44 (0) 7721 420 730
Registered in England and Wales No. 2461068 (015)
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
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 (016)