Index Cards RE: [sc34wg3] Modularization

[Bernard Vatant]

Hi all

I've kept clear from this debate for a while, out of both lack of
bandwidth, and having difficulties to really understand what the stakes
were. But reading the last posts of Lars Marius and Michel I came across
some current new (well, not that new, in fact) metaphor I have in mind
to explain topic maps. You know, that kind of marketing stuff we are all
looking for desperately ... I just want to expose here that metaphor,
and how it may help and clarify in what Michel (RM) and Lars Marius
(SAM) views are different.

In the simple cosy world of ancient (and not so ancient) libraries, you
had books and index cards. An index card for each book, with title,
author, DDC classification, format, ISBN number, date of purchase,
current status (available, borrowed by reader X) and so on. 

The book is the subject. The index card is a representation of the book
(call it topic if you like). The informations on the index card, call
them metadata, call them names, occurrences, characteristics, relations
or associations ... whatever the hell the librarian needed to know about
the book, except the book content itself, was on the index card. 
In a good library, you had also index cards for the registered readers,
for the authors, for the subjects ... When a reader borrowed a book, the
library clerck stapled the reader index card with the book index card
for a while. That was an association, right?

That is the metaphor of topic maps. Based on it, I can see three rather
independent things that we want to define and standardize.

1. SLUO : Having exact one-to-one correspondence between one index card
and one book. Or, if you have several cards, have a way to merge them.
Every good librarian knows how to do that. And you have the same
paradigm in so many places, e.g. in your adress book (one contact
person, one contact card, with whatever relevant stuff on it).

2. Assertion of relationships between cards (staples)
How do you express in a standard way relationships (associations)
between subjects, using formal binding of matching cards?

3. Structure, format, content of the cards (what information needs to be
there, under what form, what colour cards need to be, and how you manage
card boxes)

Seems to me that: 

Michel (RM) is about 1 and 2, quite mixed-up so far, and Michel thinks
now it should be separately about 1, and about 2, in two separates
modules. RM does not care at all about 3.

Lars Marius (SAM), like ISO 13250, and like XTM, tries to be about 1,2
and 3, all packed-up, and does not care to make them distinct because he
considers, as ISO 13250 and XTM have, that those three aspects are
unseparable. Take it all, or leave it all.

What Michel suggests, and I think, if I've understood well, that he is
damned right, is to make 1, 2, and 3, separate standard modules. Because
since 13250 and XTM, our reflexion has moved forward to see independent
composants where we saw a global picture at the time.

And separation between 1,2 and 3 is indeed orthogonal to current
structures of both RM and SAM. What Michel suggests, I guess, is to
restructure SAM in a way that it shows clear separation of those
aspects, so that each one could be mapped to the matching part of RM. 

Why would such modularization bring more interoperability?

Because people who achieve 1, with distinct technologies, which means
whatever the format of the cards, have a kind of interoperability
anyway. If an XTM file references a PSI URI, and an ontology in whatever
format uses this URI to define a class, well they agree on SLUO basis,
independently of whatever else. This is what people in Semantic Web are
interested in, but maybe they don't care about the rest of the package,
because they have their own way to deal with 2 and 3.

OTOH, some people maybe just want 2 and/or 3, a standard card and
relationship format for internal purposes, and do not care about SLUO
outside their intranet. Most early industrial adopters of topic maps
want exactly that, and frown if they have to buy the Semantic Web stuff
with that.

So modularization would help both internal clarification, and more
targeted marketing.

Does that help?

Bernard

_____________________________________

Bernard Vatant
Senior Consultant - Knowledge Engineering
www.mondeca.com
_____________________________________