dbedford@xxxxxxxxxxxxx wrote: (01)
>Barry,
>
>Yes. However, it is not the exact relationships that are described here that
>are as important as the idea that relationships can have meanings, and that the
>meanings can be defined and managed. This is what we are beginning to do by
>examining the relationships among concepts in large training sets which are
>domain specific.
>
>
>
This sounds right to me; however I'd like to make sure I understand what
you say correctly (see below). (02)
>I also do not think, though, that the relationships I have seen defined by the
>ontology community are sufficiently rigorous to meet the needs of individual
>domains.
>
Do you have an idea about what's missing? (03)
Since Barry made a reference to his paper, perhaps I can try to explain
what I understand is missing from current ontologies. (04)
Barry distinguishes 3 worlds of things: (05)
- concept systems (describing the meaning of terms w.r.t. relations
among terms)
- real-world entities (instances & types organized in an ontology)
- information models about the world (e.g., an allegorical document) (06)
When we use 'part-of' at the instance level to be ontologically accurate
in the sense of Barry,
we are still vague about what is the criteria by which we adjucate on
parthood. This goes
back to issues of defining criteria of identity and unity. (07)
We need identity to make sense of "a123 part-of b456". How do I
recognize 'a123' among all possible instances?
We need unity to make sense of the extent of what 'a123' refers to and
to determine whether the particular
relationship holds or not. (08)
This gets to the crux of my questions about the discussions we've had so
far:
I have a feeling that, at a coarse level, we're more or less in agreement.
At a practical level, there are wholes that are problematic w.r.t what
should we be doing next. (09)
Whether we want to use an existing ontology build our own or some
combination of the two,
we need to be practically clear about what it means to describe the
meaning of a relationship.
How many different kinds of 'worlds' do we need? thesauri, ontologies,
contexts, etc... (010)
Suppose we start with Barry's 3 worlds. It is unclear to me how these 3
are connected to one another.
At a practical level, this could create a lot of confusion w.r.t. what
these worlds are for and how we should
be using them. (011)
In this example, suppose we want to practically account for the meaning
of the 'part-of' relationship at the instance level
(i.e., the 'part-of' relationship in the ontology of real-world entities) (012)
- in 'conventional' ontologies, we'd have things like: (013)
owl:Class A
owl:Class B
part-of: A -> B
A a123
B b456
part-of(a123,b456) (014)
Is that still the right way to define the 'part-of' relationship for an
ontology of real-world entities?
I'm afraid the answer is no but I'll assume it is sufficiently
'concrete' to continue for now. (015)
How do we then explain the meaning of 'part-of'?
The ontology above certainly doesn't describe that as Denise points out. (016)
It seems to me we'd need to relate 'part-of' to things in a system of
concepts
and/or an information model. For that purpose, it seems to me that we'd need
yet another 'world': (017)
- descriptions & situations (like DOLCE's D&S ontology and the
methodology for applying D&S to a specific ontology, e.g., DOLCE-Lite) (018)
In the above example, it would make sense to associate to the 'part-of'
relationship
in the ontology of the real-world to a particular situation that
involves 'a123' and 'b456'
as an 'SItuation' instance of the 'Description' concept defining the
relationship 'part-of'
relative to specific criteria of identity, unity, dependence, rigidity,
etc... (e.g., the OntoClean metaproperties).
(I my D&S concepts aren't crossed) (019)
> The revision of Z39.19 made progress over elaborating the types
>traditional thesaurus relationships that are supported. However, the major
>area of concern in the entire approach to thesaurus relationing has for me
>always been the RT's - which is everything beyond an equivalent or a
>hierarchical concept relationship. It is in the RT areas and their
>elaboration
>that we find the convergence of ontological relationships and thesaurus
>relationships.
>
>
>
This sounds reasonable to me if a thesaurus is akin to Barry's concept
system and an ontology is about real-world entities.
Is this the sense in which you talk about ontologies and thesauri? If
so, what does 'convergence' mean?
Perhaps you mean something else entirely in which case I may have missed
your point. (020)
>Many of us have been pushing to for the critical convergence over the past
>several years. Unfortunately, there has been some resistance in both
>communities. I think the work we are trying to do at the Bank may help people
>to see the convergence. This is why in the slide I sent to NCOR I pushed for
>some real projects.
>
>
>
Is that slide available somewhere? (021)
-- Nicolas. (022)
>Best regards,
>Denise
>
>
>
>
>
> Barry Smith
>
> <phismith@buffa
>
> lo.edu>
>To
> Sent by: ONTAC-WG General Discussion
>
> ontac-forum-bou <ontac-forum@xxxxxxxxxxxxxx>
>
> nces@xxxxxxxxxx
>cc
> .net
>
>
>Subject
> RE: [ontac-forum] Surveyed Ontology
>
> 10/25/2005 "Library" Systems
>
> 12:23 PM
>
>
>
>
>
> Please respond
>
> to
>
> ONTAC-WG
>
> General
>
> Discussion
>
> <ontac-forum@co
>
> lab.cim3.net>
>
>
>
>
>
>
>
>
>
>In response to the valuable KOS document posted by Dagobert, I would
>like to draw the forum's attention to current developments on the
>ISO/CEN front regarding ambiguities in use of the word 'concept'.
>
>It is clear that the concepts represented in concepts systems stand
>in meaning relations such as narrower_than, broader_than, etc. It is
>correct, for example, that the concept structured document
>narrower_than the concept document.
>
>The problems arise with respect to
>
>3.1.3 Associative relationships
>
>for example part_of, causes, located_in, derives_from, adjacent_to, etc.
>
>It is not correct to say, for example, that the concept heart part_of
>the concept human, or that the concept cell nucleus part_of the
>concept cell, or that the concept lung located_in pleural cavity.
>
>Rather it is instances of types in reality which stand in such
>relations, and it is these relations, and the corresponding instances
>and types, which are the proper object of ontology:
>
>See http://ontology.buffalo.edu/concepts/ConceptsandOntologies.pdf
>
>BS
>
>
>
>On a proposal
>
>At 12:07 PM 10/24/2005, you wrote:
>
>
>>We do not have the resources to create an ontology library
>>system. Rather we should adopt one once we are further along. At
>>that point we can use the criteria in the Ding & Fensel article but
>>need to update the survey.
>>
>>Right now we need an ontology registry, which is a much more
>>light-weight proposition. Since much information for constructing
>>full-fledged rigorous ontologies can be gleaned from other types of
>>Knowledge Organization Systems (KOS), we actually need a KOS
>>registry. There are many attempts at creating such registries
>>(going back to the sixties), but none has been successful in the
>>sense of being maintained continuously. However, it is now possible
>>to set up a web-enabled database where authors (individuals or
>>organizations) can register their own ontologies and other KOS in a
>>structured format. (A WIKI is not the best tool for this.) Such a
>>registry should also cover use cases for ontologies
>>
>>I am working on possibly using a database tool for this purpose that
>>was prototyped by a government agency.
>>
>>Attached are two documents, one a set of evaluation criteria for KOS
>>and one giving templates for describing KOS and KOS use cases in a
>>database. This materials have been developed with more traditional
>>schemes such as MeSH or Snomed or the Art and Architecture Thesaurus
>>in mind, so they need to be extended to capture characteristics and
>>uses of formal ontologies.
>>
>>In the registry, the KOS must be identified by subject domain. Many
>>different approaches to this can coexist, and the scheme that Roy
>>suggests can certainly be one of these approaches. The concepts to
>>be used for this subject indexing of KOS need to be understandable
>>for people but need not to be as carefully specified as concepts in
>>a formal ontology.
>>
>>We formed a subgroup to consider registries. It seems that we need
>>to establish for sure who wanted to participate. Pat and I believe
>>the group includes at least the people listed below, but we also
>>believe there were more. So please add your name if you are
>>interested. The group should meet soon to work on the problems
>>outlined above and get this going, as a registry is a step that
>>should logically precede working on comparing ontologies.
>>
>>DS
>>
>>Ontology/KOS registry WG
>>
>>Pat Cassidy
>>Roy Roebuck
>>Olivier Bodenreider
>>Dagobert Soergel
>>
>>
>>
>>
>>
>>At 10/22/2005 01:18 PM, Roy Roebuck wrote:
>>
>>
>>>Content-class: urn:content-classes:message
>>>Content-Type: multipart/alternative;
>>> boundary="----_=_NextPart_001_01C5D72C.9A2912C9"
>>>
>>>Excellent! I've reviewed the material Gary cites, and agree that
>>>an "ontology library" capability would be very useful in the COSMO,
>>>ONTAC, SICOP, and Web-Service collaborations. I also submit that a
>>>"natural" outline of ontologies (i.e., packages of functions,
>>>processes, and process input/control/output/mechanism resources
>>>such as metadata, data, funds, skills) as services could be
>>>organized using the General Ontology (GO) as outlined below:
>>>
>>>
>>>
>>Dagobert Soergel
>>College of Information Studies
>>University of Maryland
>>4105 Hornbake Library
>>College Park, MD 20742-4345
>>Office: 301-405-2037 Home: 703-823-2840 Mobile: 703-585-2840
>>OFax: 301-314-9145 HFax: 703-823-6427
>>dsoergel@xxxxxxx www.dsoergel.com
>>
>>
>>
>>
>
> (023)
_________________________________________________________________
Message Archives: http://colab.cim3.net/forum/ontac-forum/
To Post: mailto:ontac-forum@xxxxxxxxxxxxxx
Subscribe/Unsubscribe/Config:
http://colab.cim3.net/mailman/listinfo/ontac-forum/
Shared Files: http://colab.cim3.net/file/work/SICoP/ontac/
Community Wiki:
http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/OntologyTaxonomyCoordinatingWG (024)
|