[Top] [All Lists]

RE: [soa-rm] RE: [ontolog-forum] RE: [soa-rm] latest Draft ofConcept Ma

To: "[ontolog-forum] " <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "Roy Roebuck" <Roy.Roebuck@xxxxxxxxxxxxx>
Date: Fri, 30 Dec 2005 14:47:01 -0500
Message-id: <878871F15E22CF4FA0CCFDD27A763B2F4CCC61@xxxxxxxxxxxxxxxxxxxxx>
Hi:    (01)

I concur with Evan's more precise explanation of RDF.  He has the deep
expertise in this area.  I have made a point of stating to this forum
that I am a technical layman, as a business manager and analyst, in
these XML-derived areas.  My observation in the cited message was
intended to provide an analogy to compare the apparently predominate use
of Topic Maps to portray "networks of relations" while technologies like
RDF tend to focus more on characterization/classification/containment
hierarchical relations, thus the analogy of the index and
table-of-contents of a book, respectively.      (02)

>From even before the advent of XML, RDF, and then RDFS after XML Schema,
could support richly-descriptive associative object models (e.g., the
OpenGroup and then OMG/CIM Object Metaschema from the early 90's, and
concept maps).  These object metaschema could enable Categorization (is
a), Containment (has a), Sequence (precedes a), Version (was a),
Variance (is from), Equivalence (is, or is partially), and Reference
(refers to) types of relations.  But RDF and RDFS are apparently seldom
used, or envisioned for use, in relation-types other than the
fundamental relation-types of Contaiment and Categorization.  This is
one reason, years ago, I moved my Catalog and Cross-Reference EA
approach onto ontology management technology rather than
RDF/metadata/MOF repository and UML/XMI/MOF technology.  This is similar
to the reason I moved away from using LDAP, in the late 90's, to support
my enterprise management approach - I couldn't find vendors and
practitioners, at the time, who envisioned LDAP as anything but an
electronic phone book.  It's also the reason I moved away from using
Project Management software to create and manage these rich object
models.  In each of these cases the underlying object technology would
easily support my approach, but the vendors would not extend their
object metaschema (M3 and M2 layers) to enable it.    (03)

It's like the difference between the software tool known as "The Brain"
which presents its content as rich "networks" of information without
providing significant "tree" structures, while tools used for
"MindMapping" will provide rich decomposible tree/hierarchy structures,
in linear or circular form, but not provide any significant capability
for linking across tree branches to form a network.  In further analogy,
a computer file system provides hierarchical folders that
"characterized" their contents, and later had "shortcuts" to link across
this hierarchical structure to provide richer associative context for
the files.  Another illustration is seen in the structure of each of the
five FEA Reference Models, providing 3 levels of
hierarchical/inheritiane/taxonomic relations between their entries, with
the vaguely-defined "lines of sight" across these reference model
entries (which are really only specified when used by the Departments
and Agencies) provide the associative/contextual/ontological relations.
The blending of these hierarchical/inheritance/taxonomic and
associative/contextual/ontological relationship-capabilities is now
found in the current generation of Concept Mapping tools such as CMAP
and Inspiration.  Concept Mapping tools enable the creation of directed
labeled graphs (DLG) with any relations among nodes that you can
imagine, including the seven relation-types I describe above and the
seven "catalog/class" types I have described in previous posts to this
forum. Concept maps seem to provide a simple and solid diagrammatic and
"triples" foundation for all controlled vocabularies, taxonomies,
semantic networks (including data models), semantic models, ontologies
(and thesauri), and knowledge/intelligence bases to support the
operations of a given community.    (04)

Roy    (05)

-----Original Message-----
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of
Sent: Friday, December 30, 2005 12:56 PM
To: ontolog-forum@xxxxxxxxxxxxxxxx
Subject: RE: [soa-rm] RE: [ontolog-forum] RE: [soa-rm] latest Draft
ofConcept Map / N-ary Documents specification?    (06)

This is a reply to a message sent some time ago.  However, I think it
sufficiently important enough to clarify its content to justify a 
response despite the tardiness.    (07)

Roy Roebuck III wrote:
>I've used the following analogy when describing Topic Maps in relation
to RDF 
>and other "knowledge-management" technologies.  --- Consider an
>to be like a "book", full of information, but not well organized, not
>navigable, and not searchable.  Topic Maps provide the functional
>of a hyperlinked "index" (including concordance capabilities) to this 
>'organization as a book", while technologies like RDF provided the
>equivalent to a "Table of Contents" to the book.
>RDF focuses on hierarchy and categorization/classification/taxonomy,
>Topic Map focuses on associations/relations/networks/lattices.  I
>that OWL provides both the "Table of Contents" and the "Index" to the
>being modeled.    (08)

This characterization of RDF is likely to be misleading for most people.
Taxonomies are usually thought of as trees made up of branches which are
totally disjoint subtypes of the type node from which they branch.  RDF
and RDFS have no way of specifying such disjointness nor is it assumed
(an rdfs:Class can be a subtype of more than one other rdfs:Class and
an individual can be typed by multiple classes as well).
Furthermore, RDF is based on a graph model which supports networks and 
lattices.  RDFS has a first class meta-element for relation, called 
Property.  RDF is not about categorization but rather about description.
It is RDF that provides URIs to identify any Resource in its universe;
OWL adds expressivity primarily for describing Resource types.  It's OWL
that adds constructs which enable classification and much richer
checking to the Semantic Web "stack".    (09)

Contrary to what the above might lead you to believe, I am not a booster
for RDF.  The above characterizations are just so off the mark, I
want them to spread.    (010)

Evan K. Wallace 
 (former participant in the W3C Web Ontology working group which
  OWL and current chair of the OMG Ontology PSIG that is developing the
  Ontology Definition Metamodel specification which includes a metamodel
  for RDF(S))
Manufacturing Systems Integration Division
NIST    (011)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (012)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (013)
<Prev in Thread] Current Thread [Next in Thread>