[wg11] Re: seds against part 11:2004

[Ed Barkmeyer]

Dear Professor Shilnikov,

you wrote:

> Dear colleague!
> About this quotation:
> trc> Part 11 formally contains the concept "multi-leaf entity instance"
> trc> (3.3.11) and the notion that every such instance has a "multi-leaf
> trc> complex entity data type" (3.3.12).  Every instance of a given
> trc> multi-leaf complex entity data type must be a valid instance of every
> trc> entity data type that appears in it.  It follows that a multi-leaf
> trc> complex entity data type must in fact be a *subtype* of every entity
> trc> data type that appears in it!
>
> I agreed with this statements, but I offer to rename the concept
> "multi-leaf" or "complex" etc... to "multiple data classification", as used
> in UML world for such things.  It will be a right step to
> harmonization with UML.  As about me, I just use the term "multiple
> data classification" in my lectures for students.

Some points:

First, the ISO activity that maintains EXPRESS is TC184/SC4/WG11.  There 
are several recently published SEDS (discrepancies) in ISO 10303-11:2004 
(EXPRESS, Edition 2), and it is likely that a Technical Corrigendum will 
be produced in the near future.
   Russia is a Participating member (thru GOST), and your technical 
expertise is very welcome.  The WG11 email exploder is 
wg11 at steptools.com, and the ISO Convenor is David Loffredo at STEPTools 
(USA), loffredo at steptools.com.  (Officially, you would need to be 
identified as a participating expert by GOST, but your technical 
comments are welcome in any case.)

Second, there are two ongoing activities relating EXPRESS to UML, one in 
ISO (WG11) and one in OMG.  The ISO activity, designated ISO 10303-25 
revision, is explicitly concerned with mapping EXPRESS v2 (2004) to UML 
v2 (2004).    The lead expert on "Part 25" is David Price of Eurostep 
(UK), david.price at eurostep.com.  The OMG activity is developing a formal 
MOF (ISO 19502) metamodel of EXPRESS and a formal mapping to the UMLv2 
metamodel (among others).  The lead expert on that is Uwe Kaufmann of 
Fraunhofer IPK (Germany) -- uwe.kaufmann at ipk.fraunhofer.de, but you will 
find that most of the participants in the two activities are the same. 
Your participation in either or both of these is welcome.

Finally, the terminology issue you raised in your email -- to use 
"multiple classification" instead of "complex entity data type" or 
"multi-leaf entity data type" -- needs a bit of work.  The problem is 
that the definition of the term "multiple classification" in UML (ISO 
19501) is closely related to, but not the same as, any of the named 
EXPRESS concepts.

In EXPRESS, a "complex entity instance" is an object that is an instance 
of more than one modeled entity data type.  And a "complex entity data 
type" is a set of modeled entity data types that can be simultaneously 
instantiated in some object(s).  In EXPRESS, there is no constraint as 
to how those types may be related:  if B is a subtype of A, then every 
instance of B also instantiates A, and therefore every instance of B is 
a complex entity instance.  In UML "multiple classification" relates an 
object to a set of classifiers that are *not related by generalization*, 
and so it excludes simple subtypes.

In EXPRESS, a "multi-leaf entity data type" is a complex entity data 
type in which no one entity data type is a subtype of all the others. 
In ISO 10303-21 and -28, instances of such a data type are said to be 
"uncharacterized" -- there is no named entity data type which has or 
inherits all of their properties.  So it contains at least two "leaf" 
entity data types that taken together have or inherit all the 
properties, but are "not related by generalization".  And this is the 
same concept as UML "multiple classification".

But in ISO 19501 (UML), "multiple classification" is defined as "A 
semantic variation of generalization in which an object may belong 
directly to more than one classifier."  And "generalization" is a 
*relationship* among model elements (including the relationship between 
an object and a classifier).  So "multiple classification" is the 
relationship between a "multi-leaf entity instance" and the "leaf" 
entity data types contained in the "multi-leaf entity data type".  But 
it is not a synonym for either of the EXPRESS terms.

Further, in each case where ISO 19501 refers to a "multi-leaf entity 
instance", it uses a circumlocution such as "an object that belongs 
directly to more than one classifier ([see:] multiple classification)". 
  So UML does not have a term for either the multi-leaf instance or the 
"multi-leaf entity data type" (the set of classifiers).

So we cannot substitute "multiple classification" for either EXPRESS 
term.  But it does seem advisable to modify the definition of 
"multi-leaf entity data type" to add a note to this effect.  It would 
read something like:

"multi-leaf (complex) entity data type
"a complex entity data type in which no single entity data type is a 
subtype of all the others in the collection.

"Note  A multi-leaf entity data type does not have an identifier in the 
schema.  It describes a set of complex entity instances that have 
'multiple classification', as defined in ISO 19501:2005."

Would this be satisfactory?

Best regards,
-Ed Barkmeyer

-- 
Edward J. Barkmeyer                        Email: edbark at nist.gov
National Institute of Standards & Technology
Manufacturing Systems Integration Division
100 Bureau Drive, Stop 8264                Tel: +1 301-975-3528
Gaithersburg, MD 20899-8264                FAX: +1 301-975-4694

"The opinions expressed above do not reflect consensus of NIST,
  and have not been reviewed by any Government authority."