W3C home > Mailing lists > Public > public-owl-comments@w3.org > January 2009

Suggestions to improve the introduction to 'Structural Specification'

From: Jonathan Rees <jar@creativecommons.org>
Date: Fri, 23 Jan 2009 20:38:05 -0500
Message-ID: <760bcb2a0901231738o1e65c88dicc1142919716ea92@mail.gmail.com>
To: public-owl-comments@w3.org

I, too, will send my comments along in separate messages.  Here is the
first.  (Some or all of the rest will probably miss your January 23
deadline, sorry.)

I misread the following text in the SSFSS introduction, and fear that
others will be confused by it too:

   OWL 2 ontologies consist of the following three different syntactic
   * Entities, such as classes, properties, and individuals, are
   identified by IRIs and can be thought of as primitive terms or names.

I read this as saying that the category of entities is one of those
three syntactic categories.  This would imply that classes,
properties, and individuals are syntactic in nature.  I really thought
for a minute that you were going down this road (term models?), and it
took me a while to see that there was another way to read what
was said.

My guess is that you meant for this sentence to be read as saying that
the first syntactic category is that of IRIs, which identify entities,
and that these IRIs "can be thought of" primitive terms or names.
Here's my attempt at a rewording:

   OWL 2 ontologies are composed from ['built from', 'use', etc.] the
   following three syntactic categories:
   * Entity names.  Entities such as classes, properties, and
   individuals are named by IRIs, and their names form the primitive
   terms of the ontology.

I don't understand "can be thought of" - the entity names *are* the
primitive terms in the ontology, aren't they?

I think there is an unstated idea that might be worth drawing out in
the introduction, that the ontology *specifies* a model (or more
accurately: specifies nonuniquely what it means for a given thing to
*be* a model, but that can wait until the model theory document), and
the entities in any model are meant to model things in the domain,
such as people, sets of people, relationships between people, etc.
Because this is not stated, I think it will be easy for readers to get
confused between the ontology (which is syntactic, right?) and the
model(s) (not syntactic).  In particular, entities belong to the model
(or a model), not to the ontology, while names and expression are used
in the ontology, but don't exist in the model. The ontology describes the
model, and the model models the domain.

One way to bring this out would be for the introduction to provide a
definition of "ontology" consistent with the way the word is used in
all the documents.  Perhaps this is the same as what section 3 says,
but I think the introduction could say the same thing less technically
than section 3 does, while being a bit more specific than "formal
conceptualization of a domain of interest".  I'm thinking something
like "formal description of a model [or 'conceptual model'] of a
domain of interest" but I admit that's kind of awkward.

Am I right that you use "model" and "represent" intechangeably, and
that this is a relation between entities in a model and things, sets,
or relationships in the domain of interest?  It might be helpful to
use the same word consistently, and explain it.

Maybe I'm being pedantic, but this stuff is very confusing to the
uninitiated and anything we can do to steer people straight will pay

Received on Saturday, 24 January 2009 01:38:44 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:01:29 UTC