[sc34wg3] Some general comments on the RM (branching from the
thread Re: [s c34wg3] The Norwegian National Body position on ISO 13250)
Mason, James David (MXM)
Tue, 15 Apr 2003 17:30:57 -0400
It's interesting to see you agree that the RM is syntactic at the same time
that I'm getting other people wondering how it can possibly be syntactic.
I agree that the SAM has much more concrete semantics because it develops
things like names and associations. But the RM has to establish some
foundation for your ability to create those. Is a name a type of assertion?
I think so. Is an association also an assertion? Certainly. Figure 1 in the
current RM started out, I believe, as a detailed dissection of an
association. The current figure simply removes the concrete subjects (which
were orignially things like "Lena Horne" and "Stormy Weather" and later
became someone with a Harvard M.D.) and leaves you with bones and no skin.
It's OK for the SAM to start putting skin on the bones, but the RM is so
abstract now that the wind will blow those dry bones away.
From: Lars Marius Garshol [mailto:firstname.lastname@example.org]
Sent: Tuesday, April 15, 2003 5:06 PM
Subject: Re: [sc34wg3] Some general comments on the RM (branching from
the thread Re: [s c34wg3] The Norwegian National Body position on ISO
* James David Mason
| I observed that I felt that the RM was the place for fundemental
| semantic grounding for TMs. At this point, the RM is still heavily
| syntactic. It places a lot of emphasis on properties (Clause 3) and
| constraints (Clause 4). This material, particularly in Clause 4, is
| syntactic, though at a very abstract level. Perhaps I should say
| it's structural, but for me the listing of properties comes closer
| to syntax than just structure. That syntactic material is necessary
| to provide a basis for mapping to the SAM. But it's severly lacking
| when it comes to defining the semantics of TMs.
I think you are right in saying that the RM is rather syntactical,
though "abstract syntax" would probably be the most correct.
However, I'm not sure this is a fair criticism. The RM consists of
things and statements about things (topics and assertions) and there's
not a whole lot of semantics you can put into that, simply because it
is so basic.
There are two more reasons why I am concerned that this may not be
a) only the people who created the thing can really claim to decide
what it is supposed to do. I have said many times, and will
probably repeat many more times, that I don't know what it is
for. I'm hoping the authors will explain it to me one day, but so
far I, personally, don't have an understanding of what this thing
was created to achieve. Until I do I feel the only grounds on
which I can criticise it are the purely esthetical. Which may be
useful, but only on the microlevel.
b) the RM can't really describe what base names, occurrences, scope,
and variant names are, as these things do not exist in the RM. So
if you want to have the semantics of these things it just doesn't
work to look for them in the RM. (Whether that's what you're
asking for I don't know, but since nobody else is speaking up for
the thing I am trying to put a possible alternate view.)
| In short: You need to be standardizing both syntax and semantics; at
| this point you're neglecting semantics. Syntax/structure should be
| defined as concisely as possible, preferably through notation or
| tables rather than prose. Prose should be reserved for things that
| can't be done in notation, which means semantics.
The SAM does attempt to define some semantics in prose, such as
"An occurrence is a relationship between a subject and an
information resource. The subject in question is that represented by
the topic in whose [occurrences] property the occurrence item can be
found. The precise nature of the relationship is described by the
occurrence type, a subject which is attached to the
occurrence. Occurrences are generally used to attach information
resources to the subjects they are relevant to. [...]
Note that the occurrence is properly not the resource, but the
relationship between it and the subject. Occurrences are essentially
a specialized kind of association, where one participant in the
association must be an information resource."
I'm not sure that it is possible to do much more than this, and
certainly I am baffled as to how it should be done. I
think it was the idea behind Daniel Rivers-Moore's proposal for a
conceptual model to do more, but it would take a better mind than mine
to do so.
Semantics are notoriously slippery and any attempt to pin them down
too much will be met with "but my application needs to do X", or "I've
seen it used like this", and ... And quite reasonably so.
Lars Marius Garshol, Ontopian <URL: http://www.ontopia.net >
GSM: +47 98 21 55 50 <URL: http://www.garshol.priv.no >
sc34wg3 mailing list