Cory - Pitching in, but really with no time to do this. So brief I will be.
Cory Casanave wrote:
Hi Dean,
Yes, skos:exactMatch is much more palatable. However I wondering if it
may be a good idea to more formally recognize this difference between an
information resource and what the information resource is about (the
thing in the world).
When we want to understand there are different viewpoints and
authorities making statements about something then we start to want the
information to have its own identity and to be able to talk about that
information as different from the subject of the information.
Any ontology work must recognize that there are different spaces (call them Information Spaces - IS) in which meanings are shared. In the case of an IR, in one IS it might have "about-hood" in another space it may be "of-hood". What does this mean? The notion of "thing-in-the-world" (TINW) should be carefully discerned when we are talking IRs.
I will use examples from System Engineering, SE, (not Software Engineering - hard sometimes to agree "TINW" for that often). So if I have an ORION vehicle (future APOLLO-like space vehicle) what am I referring to - an instance of an ORION, a type of vehicle, the "As-Designed", "As-Built", "As-Maintained" IRs? And what Aspect and Viewpoint are we dealing with (notice I distinguish "aspect" from "viewpoint"). IRs exist for different aspects of the vehicle and different viewpoints, and an IR has "about-ness" that represents a transference of concepts, properties and axioms to a specific IS - the map of the Mall is not the Mall.
One further piece, an IR can also have an IRD - an Information Resource Description - what an IR is. An example helps. Consider an Interface Control Document (ICD) as an IR. There is a ICD Description saying that an ICD has two or more system elements and a set of constraints for the interfacing, etc. Condensing 3 modeling styles into one sentence: this would be an owl:Class (or sub-class of owl:Class if OWL-full is allowed) with sub-classes (or owl:Classes) for different types of ICD (or instances if expressing types as instances is acceptable practice ). A particular ICD for a specific type of interfacing would reference the ICD Description. Sometimes this is referred to as a "class-instance" mirror (or metaclass-class mirror) pattern.
Where do we find SE models that make these distinctions? As you may know, I have taken SysML and made OWL versions in SysMO. DoDAF 2.0 we have converted to OWL. NIEM also. All these offer examples of IRs, IRDs and ISs. At NASA, NExIOM ontologies have these distinctions because Enterprise Architecture meets Data Architecture and System (think vehicle as as well as Information System) Models.
For oeGov, we have a start on separating these concerns. If we are to collaborate on this, as you well know, joint work needs a common meta-level of how we want to express our cognitions about "things" and "spaces". If we achieve that we progress at the model level. Currently LOD is happening with "little-o" efforts - very light ontologies. The pragmatists of LOD are doing quite well without heavy models. So perhaps there is time to get "midsize-o" work happening?
This gets
back to some of the reification issues we discussed long ago.
So if we wanted to say that <information resource> represents <identity>
we could have the option of combining or contrasting information
resources about the same identity. The other way may be to say that
<information resource> skos:exactMatch <information resource>, but that
is a somewhat different semantic.
Meanings are affordances that happen when we are clear on how cognitions in one space of viewing the world are applicable in other (information) spaces. These spaces constitute how we organize our representations for putting distinctions to work - we might think of this as "transference of meanings". Simply stated - a model is a view for a purpose (how 'applicable' has meaning). There is difficulty always in thinking about notions of the "real world".
I just purchased you book and have not had a chance to read it yet -
perhaps it has all the answers!
It would seem to me that some consensus on such an approach for LOD
would be a good idea. Thanks for the pointers.
-Cory
-----Original Message-----
From: semweb-31@xxxxxxxxxx [mailto:semweb-31@xxxxxxxxxx] On Behalf Of
Dean Allemang
Sent: Wednesday, September 02, 2009 5:42 PM
To: semweb-31@xxxxxxxxxx
Subject: Re: [semweb-31] Linked Open Data Questions
Cory Casanave wrote:
We distinguish "information" resources from logical ones.
Informational resources are dereferenceable as a URL and have a set of
triples that are the facts about that informational resource from the
point of view of the publisher, who is the owner of that URL.
Informational resources have a triple (not yet sure if we want to commit
to sameas) that references the identity of the logical resource that is
the subject of that informational resource. Logical URI is the identity
of the real world entity or concept, not information about that entity
or concept. So the information has an identity that is connected to but
not the same as the underlying entity.
Just on the topic of commitment to owl:sameAs, an alternative (which is
still a commitment, but perhaps more palatable in some contexts) is to
commit to skos:exactMatch
(http://www.w3.org/TR/2009/PR-skos-reference-20090615/#mapping). I have
found this to be sensible in certain contexts where, for one reason or
another, I am not quite willing to make the very strong logical
commitment that owl:sameAs entails.
Dean
--
Please Note: If you hit "REPLY", your message will be sent to everyone
on this mailing list (semweb-31@xxxxxxxxxx)
http://semweb.meetup.com/31/
This message was sent by Dean Allemang (dallemang@xxxxxxxxxxxxxxx) from
The Washington Semantic Web Meetup.
To learn more about Dean Allemang, visit his/her member profile:
http://semweb.meetup.com/31/members/8494388/
To unsubscribe or to update your mailing list settings, click here:
http://www.meetup.com/account/comm/
Meetup Support: support@xxxxxxxxxx
632 Broadway, New York, NY 10012 USA
--
Please Note: If you hit "REPLY", your message will be sent to everyone on this mailing list (semweb-31@xxxxxxxxxx)
http://semweb.meetup.com/31/
This message was sent by Cory Casanave (cory-c@xxxxxxxxxxxxxxx) from The Washington Semantic Web Meetup.
To learn more about Cory Casanave, visit his/her member profile: http://semweb.meetup.com/31/members/9543609/
To unsubscribe or to update your mailing list settings, click here: http://www.meetup.com/account/comm/
Meetup Support: support@xxxxxxxxxx
632 Broadway, New York, NY 10012 USA
--
Ralph Hodgson
CTO and Executive Partner, TopQuadrant, Inc., www.topquadrant.com, http://twitter.com/TopQuadrant
VA Office: (703) 299-9330, CA Office: (650) 265-0529, eFax: (425) 955-5469, Cell: (781) 789-1664
http://twitter.com/ralphtq
--
Please Note: If you hit "REPLY", your message will be sent to everyone on this mailing list (semweb-31@xxxxxxxxxx)
This message was sent by Ralph Hodgson (rhodgson@xxxxxxxxxxxxxxx) from The Washington Semantic Web Meetup.
To learn more about Ralph Hodgson, visit his/her member profile
To unsubscribe or to update your mailing list settings, click here
Meetup Support: support@xxxxxxxxxx
632 Broadway, New York, NY 10012 USA