[xtm-wg] Whatever happened to facets?

[Steve Pepper <pepper@ontopia.net>]

At 17:49 29.04.2001 -0400, Thomas B. Passin wrote:
>> 6) Topic Maps (using a unary association) [1]
>>
>> "There is a topic with the ID 'shoe-257' of type 'shoe',
>> and an association of type 'is-red' in which
>> 'shoe-257' plays the role 'object'.
>>
>
>So that's what happened to facets!  They were replaced by unary
>associations.  I always wondered.

Well, no, not quite...

<INTRO PURPOSE="to get everyone on the same page">

Facets are a concept found in ISO 13250, the original topic map
specification. They are not in XTM, despite the stated goal that
"XTM shall be compatible with ... ISO 13250".

13250 has the following definitions:

   3.9 facet
     a) The subset of information objects that share an
        externally-applied property.
     b) The values given to a particular property externally
        applied to a set of information objects.

   3.10 facet link
     A hyperlink that applies values for a given property
     (as well as the property itself) to one or more
     information objects.

  3.11 facet type
     A property applied by one or more facet links to one or
     more objects.

   3.12 facet value
     A member of the set of all values of a particular facet type.

See http://www.y12.doe.gov/sgml/sc34/document/0129.pdf

Basically, facets allow you to specify property-value pairs for
resources that are occurrences of topics. 

[Why would you want to do this, instead of scoping the relevant
occurrence? Because many properties, e.g. language, are properties
of the *resource itself* rather than its use *as an occurrence*
of a particular topic. The same resource may be an occurrence of
many topics; the *scope* of those occurrences might differ, but a
property such as language will always be the same.]

</INTRO>

During development of 13250 several key people were vehemently
opposed to including facets in the standard, because it was felt
that they were orthogonal to the topic map model itself. Others
felt that they were such an extremely useful adjunct to that
model in real life applications, that they should be included,
and in the end they were.

After the event, it turned out that facets were underspecified,
and that there were a number of misconceptions about how they
should be used. In particular, there was a lot of misuse of the
[facetval] attribute in the <fvalue> element. Once this was
cleared up, it was pointed out that having the declared value
of [facetval] to be NAME essentially made facets useless in
many situations. (It meant you couldn't have the value of a
facet be, say, "42", unless you created a topic solely for that
purpose.)

So there was already considerable disgruntlement with facets
by the time XTM was put on the agenda.

During the development of XTM, a recurring theme was the
relationship between topic maps and RDF, and one of the points
made repeatedly was that anything facets could do, RDF could do
better: after all, RDF is a framework for applying metadata,
i.e. property-value pairs, to resources. (I'm not sure that this
is entirely true, at least until such time as RDF and topic
maps are aligned in such a way that they can be used in tandem,
but the claim was made.)

Also as a result of trying to understand what RDF really was
all about, and how it fitted with topic maps, the concepts of
reification and of addressable and non-addressable subjects
were developed.

Now, since a resource *is* an addressable subject, it is easy
to reify it as a topic. Once that is done, it can be assigned
characteristics, including property-value pairs. This could be
done either through occurrences or associations (which amount
to basically the same thing, since another insight of the XTM
process revealed that occurrences were in fact just a special
kind of association).

The details of the mapping between 13250 facets and XTM
constructs has not been worked out in detail, and certainly
not specified anywhere. (I believe the foregoing explanation
is the first time anyone has actually written up this stuff...)
This needs to be done and it is one of the aspects of the
relationship between 13250 and XTM that will be addressed at
the upcoming SC34 meeting in Berlin.

So, Thomas, you were close. Facets *did* get replaced by
associations, but those associations need not be unary, and
they could also be occurrences.

Hope this helps.

Steve

--
Steve Pepper, Chief Technology Officer <pepper@ontopia.net>
Convenor, ISO/IEC JTC1/SC34/WG3  Editor, XTM (XML Topic Maps)
Ontopia AS, Waldemar Thranes gt. 98, N-0175 Oslo, Norway.
http://www.ontopia.net/ phone: +47-23233080 GSM: +47-90827246


To Post a message, send it to:   xtm-wg@eGroups.com

To Unsubscribe, send a blank message to: xtm-wg-unsubscribe@eGroups.com 

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/