[Top] [All Lists]

Re: [ontolog-forum] Constructs, primitives, terms

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: William Frank <williamf.frank@xxxxxxxxx>
Date: Sat, 17 Mar 2012 15:08:46 -0400
Message-id: <CALuUwtBBBXb9VARZZqPodqF+hygfMyXx2eWhygupWyuanV9Wbg@xxxxxxxxxxxxxx>

On Sat, Mar 17, 2012 at 11:33 AM, John F. Sowa <sowa@xxxxxxxxxxx> wrote:
On 3/17/2012 9:38 AM, William Frank wrote:
> while the official semantics of UML make that UML semantics a dead end,
> hanging off the end of an aging programming style.

UML is a collection of diagram types that are used for various aspects
of semantics. There is no reason why anybody has to use all of them, and there
is no reason why more (or different) diagram types can't be added.

This is exactly what I think UML *should* be, and ***can be** used as. It *should be regarded as, and is actually uses as* a collection of diagram types, that can and are extended (and repurposed as needed, like any language). I really was so sure that this is what I said in my message.

Just look at three of the basic diagrams:

I do, every day, and create them, not just look at them. Really really good ones.

1. The class diagrams are based on the Tree of Porphyry, which
was introduced about 1800 years ago to represent Aristotle's
categories. It represents the backbone of any ontology,
including anything represented in OWL.

Yes, this is what I think they SHOULD be. Unfortunately, if one is willing to waste one's time reading the official semantics, which I believe only 1 in 100,000 would, one finds that UML classes are not and can't be categories of things, in the official semantics, they are a *conflation* of templates that can be used to produce things (like software classes in class oriented programming languages and specifications of categories of things, where UML does not distinquish between a specication and the category so specified (which is the equivelence class of all the things that satisfy the specification.)
2. The Entity-Relationship diagrams are based on Bachman diagrams
and Chen's E-R diagrams from the 1960s and '70s. They represent
signatures and cardinality constraints that are useful for any
version of logic, including OWL.

3. The component diagrams represent part-whole relations that are
also represented in any ontology language, including OWL.
Those three diagrams alone cover 99% of the OWL ontologies. Translating
them to any version of logic (including OWL) is trivial.

Absolutely!!! That is the very point.

> ... being academic sticklers for every word everyone uses, (important
> in standards documents, but in emails?), instead of taking a generous
> view of others' intents, and the arrogance of standards bodies, has
> brought us to this tragic state of affairs.

What tragic state of affairs?

That the people in computer science don't universally use UML diagrams for conceptual modelling, as the way of communicating with *each other*. Instead, they use different languages for different ways of talking to their computers.  The h with each other.  But a unified modelling language is what *should* have happended. As you say, it might not be too late.

As I said, am an advocate and very long term practitioner of using UML for conceptual modelling. Only since Aristotle, few have beleieved that classes come first, and the things are just their reflexions on the cave wall, the way official UML does.

As I said in my message, it is an almost impossible natural language that does not essentially expect things and their types to be almost always spoken about in the very same sentences, as logics do. A language with an underlying ontology in its official semantics that uses a ***different name*** for a thing of a type and a type (dog and droogle in my example - I saw a droggle today, that is, a member of the clas "dog") would be almost impossible to learn. Fortunately, no actual users of UML bother with this silliness, and call each individual message a message, not a signal, when they speak, etc. etc.

But, the arrogance of the fathers of UML, in not checking in with the 2500 years of logic you refer to in creating their own "semantics" caused part of this problem. When people want to do everything again every time, as for some reason they do in IT, they find an excuse when they read the official semantics. Just so, W3C decided to use standard logic concepts, just so, did the creators of component architecture technologies.

It is also a tragedy of the kiind I talk about in my message that you would think and even assume that I would opine about UML in a public forum without knowing a fair bit about it. I was very active in the creation of UML2, and argued strongly and unsucessfully that its semantics should be based on standard model-theoretic ideas.

I see, though, that you cut off the rest of my message, where I explained the points I made in my summary at the top.

My recommendation is to look at what
Google, Apple, and IBM have been doing. Their profits are soaring.

The WWW is big, but the SW is tiny. And the big boys make money
without using the SW tools. They must be doing something right.

Agree wholeheartedly!

But just think, if we could have a single conceptual modelling technique that was used both by web developers and others.


William Frank


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)

<Prev in Thread] Current Thread [Next in Thread>