[sc34wg3] Draft Reference Model

Lars Marius Garshol sc34wg3@isotopicmaps.org
23 Nov 2002 11:15:10 +0100

* Sam Hunting
| Well, an assertion is not homeomorphic to an association, in the
| sense that there is not a one-to-one correspondence between markup
| constructs (information items?) and graph constructs.

OK, that seems to confirm my understanding of this. 
* Lars Marius Garshol
|  b) their structure is normalized so that the original <member>
|     structure is entirely lost.
* Sam Hunting
| Presumably a graph could be constructed, in some application
| definition that preserved this information. But there seemed to be
| no reason.

I don't think there is any reason to keep it, either. I was just
checking that I'd understood correctly. The whole point of having a
model is to make it clear what information should be lost.
| I'm not entirely comfortable with the word "normalize" -- on the one
| hand, normalization has connotations of simplicity, which is what we
| are after (fewer components); on the other hand, in the database
| world as I understand it, a relation is by definition normalized,
| and the kind of "normalization" we are talking about here (not
| reflecting the <member> element in the graph) is a function of
| application definition, not a feature of the RM as such. So I think
| introducing the word "normalize" would be confusing.

Just about any word would have other uses in other communities, and
the fact that a word is used in another community does not
automatically make it confusing. Think "normalized" as in "normalized
SGML", and you've pretty much got it.

I think normalization is likely to become quite important for topic
maps in the future, because if you want to work on an XTM document
with XSLT or XML Query you have to assume that it has been normalized
first. (What Jonathan Robie & I called "cooked" in "The Syntactic

Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
ISO SC34/WG3, OASIS GeoLang TC        <URL: http://www.garshol.priv.no >