LANG: Summary of Issues 5.6 and 5.14

Guus has asked me to provide a summary of the discussion on issues 5.6
and 5.14. Hopefully, this summary will help us see what the major
questions are and help us to decide how to proceed.

ISSUE 5.6: owl:imports
-------------------------
This has generated quite a lot of discussion, and there seem to be a
number of questions that must be answered.

1) How do we define the meaning of owl:imports?

Alternatives:

a) An entailment based approach proposed by me [1]. In short, the triple
A owl:imports B means if graph(B) entails X then graph(A) entails X

Pat Hayes suggested something similar in [2]: "If an ontology A contains 
[import B] (in whatever notation turns out to be appropriate) and if 
B + A entails C then A entails C"

b) Peter's approach from [3] where inclusion occurs at the abstract
syntax level. I think here you would translate each ontology to the
abstract syntax and then merge them together before applying the
semantics.

c) Parse each ontology into its RDF graph, merge the graphs, and then
apply the semantics to the merged graphs. I don't think I've heard
anyone advocate this approach yet.

d) Do inclusion at the RDF/XML syntax level. Methods for doing this have
been proposed by Mike Smith [4] and Raphael Volz [5].


2) Are owl:Ontology and owl:imports in the domain of discourse? That is,
what kinds of statements can we make about ontologies and imports
statements?

Alternatives:
a) They are just syntax outside of RDF. You cannot make OWL or RDF
statements about them

b) They are expressed in RDF but are "dark" to OWL. That is, they are in
RDF's domain of discourse but not in OWL's, which is a subset of RDF's
(along the lines of Pat's proposed semantics)

c) They are in both the RDF and OWL domains of discourse.


3) Is it important to preserve the relationship between statements and
the ontologies from which they come?

This is more of a YES or NO question, rather than a set of alternatives.
The answer may influence the choice of how we define owl:imports 


4) What is the meaning of referencing a class or property without
importing the ontology from which it comes?

Alternatives:
a) The importing ontology is considered ill-formed

b) The term is used but none of the statements from its source ontology
contribute meaning in the importing ontology. This is the position
advocated by Peter in [6].

c) The usage imports some, but not all of the statements from the source
ontology. I believe this is the approach advocated by Jim in [7].

d) The usuage implicitly imports all of the statements from the source
ontology. I believe this is advocated by Chris in [8]



ISSUE 5.14: Ontology Versioning
--------------------------------
The original proposal is [9]. This hasn't generated nearly as much
discussion as issue 5.6.

- Mike Dean suggested some additional RDF contstraints on the versioning
vocabulary [10]

- Peter has expressed some concern whether owl:Ontology is in the domain
of discourse. This is related to question 2 from the previous issue

- How does backward compatibility relate to imports? Does it change its
meaning? Does it automatically import the earlier version of the
ontology?



I hope this summary captures the gist of the dialogue. There's been a
lot of discussion on these topics, and there's a chance that I missed
something or misunderstood someone's position. I ask that those of you
who have participated in the conversation carefully read this message
and see if I have accurately potrayed the various issues and
alternatives. If I missed anything, please bring it up.

Jeff

[1] http://lists.w3.org/Archives/Public/www-webont-wg/2002Sep/0089.html
[2] http://lists.w3.org/Archives/Public/www-webont-wg/2002Sep/0218.html
[3] http://lists.w3.org/Archives/Public/www-webont-wg/2002Sep/0216.html
[4] http://lists.w3.org/Archives/Public/www-webont-wg/2002Sep/0161.html
[5] http://lists.w3.org/Archives/Public/www-webont-wg/2002Sep/0227.html
[6] http://lists.w3.org/Archives/Public/www-webont-wg/2002Sep/0259.html
[7] http://lists.w3.org/Archives/Public/www-webont-wg/2002Sep/0225.html
[8] http://lists.w3.org/Archives/Public/www-webont-wg/2002Sep/0231.html
[9] http://lists.w3.org/Archives/Public/www-webont-wg/2002Sep/0090.html
[10] http://lists.w3.org/Archives/Public/www-webont-wg/2002Sep/0094.html

Received on Thursday, 19 September 2002 10:51:21 UTC