Page 1 of 1

Some questions on Geometry Model Clarification document

Posted: Sat Nov 10, 2012 2:01 pm
by vvagr
Questions to http://www.linkedin.com/redirect?url=ht ... cking_anet

1. There is probable a mistake in:
Notation
- If a word appears in underlined, bold camel case in green it refers to ISO15926 Part 4.
Should it be Part 3?

2. Why Part 3 entities are refereed to as Classes? In line with Part 7 they should be called "entity types" or unary first-order predicates in FOL template description language.

In current template definitions statements like
ClassOfAbstractObject(Class BoundedFiniteSurface) ^
UnboundedFiniteCurve ( x1 ) ^
are quite clear but formally meaningless.

3. Related to that, can you clarify the problem identified as:

"There is ongoing discussion about :-
o How to define a single template to achieve identification and classification of abstract objects"


and

"These issues are realized by the omission of the role hasPossessor/hasObject ".

What is wrong with adding hasPossessor x0 role to each template and start the axiom with proper typing predicate?

Like instead of:
BoundedFiniteSurface ( x1, x2 ) <->
ClassOfAbstractObject(Class BoundedFiniteSurface) ^
UnboundedFiniteCurve ( x1 ) ^
FiniteSetOfDirectedLoop ( x2 )
put:
BoundedFiniteSurface ( x0, x1, x2 ) <->
BoundedFiniteSurface(x0) ^
UnboundedFiniteCurve ( x1 ) ^
FiniteSetOfDirectedLoop ( x2 )

Re: Some questions on Geometry Model Clarification document

Posted: Thu Nov 15, 2012 4:21 pm
by dariuskanga
Victor, thanks for your comments. Here are the Geometry SIG team's preliminary responses.

1) We have revised the notes in question as follows to be more clear.
- If a word appears in underlined, bold camel case in green it is an existing class in the JORD RDL.
- If a word appears in bold camel case in orange it is a new class not currently in the JORD RDL.

2) We are following up with the Part 3 author on the question of Part 3 entities types.
Can you please clarify your comment on current template definitions statements or suggest how to make them formally meaningful?

3) As stated in the Issues section of the document, the omission of the "hasPossessor" role relates to the issue of persistent identity and will be discussed separately.

Re: Some questions on Geometry Model Clarification document

Posted: Thu Nov 15, 2012 4:38 pm
by vvagr
Thank you for answers.

I can not decipher what the term ClassOfAbstractObject(Class BoundedFiniteSurface) may mean in the FOL language of Part 7.

I'm calling "formally meaningful" exactly the form I'm suggesting in my question 3 - BoundedFiniteSurface(x0)

Please clarify what exactly is the issue of persistent identity you are reffering to?

Re: Some questions on Geometry Model Clarification document

Posted: Mon Nov 19, 2012 6:56 pm
by dariuskanga
Thanks Victor. The SIG is not meeting this week due to holidays, but it will discuss and respond at the next SIG meeting.

Re: Some questions on Geometry Model Clarification document

Posted: Thu Dec 06, 2012 3:43 pm
by dariuskanga
Victor,
We have discussed your comment in today's SIG meeting.

The intention is to have a single template that represents an Abstract Object (such as a BoundedFiniteSurface).
The statement ClassOfAbstractObject(Class BoundedFiniteSurface) is intended to classify the abstract object represented by the template as a BoundedFiniteSurface.

We want to avoid having a persistent identifier for all geometric objects to eliminate the lifecycle and maintenance issues. These are necessary for business objects but not geometric primitives. This is why we want to represent an abstract object as a template.

Can you suggest a way to accomplish this?

Thanks,
Geometry SIG

Re: Some questions on Geometry Model Clarification document

Posted: Thu Dec 06, 2012 5:15 pm
by vvagr
Colleagues,

As far as I understand, there are no differences between persistent and temporal objects on the technology layer, be it FOL representation or RDF/OWL serialization. Take a look at the Ur-Class methodology http://15926.org/publications/general-d ... /index.htm suggested by Hans. There are both persistent Ur-Classes and component variant-classes of "temporal" nature, created "on the fly". But there are no differences in implementation - both have URIs, both occupy template roles described in accordance with Parts7,8.

Lifecycle maintenance is a process, if you don't need it - you just do not bother about keeping URIs permanently and throw out your data as soon as handover is completed. But if you want to bring Geometry data to the happy new RDF world - you need URI identifiers, and you need template role for "object in focus" in each and every template.

You can parse set of "quasy-template" statements you are suggesting and recreate objects, no doubt about that. But this implementation will be different from proper Part 8 implementations, and mapping methodologies and tools will not work for geometry data.

Don't worry, there are plenty of unique UUIDs available, and generation algorithm is very robust :-)

Re: Some questions on Geometry Model Clarification document

Posted: Thu Jan 10, 2013 3:45 pm
by dariuskanga
Victor,
The SIG will be revising the Geometry Model Clarification document next week to address the issue you raise.