[Sigia-l] multi-faceted & topic map

[Lars Marius Garshol]

* Peter VanDijck
| 
| The tide is turning :)

Let's hope so. :-)
 
| Almost any classification structure can be expressed as an XML
| topicmap.  You can also express almost any classification structure
| as RDF for example, another format for metadata. The question is why
| do you want to do that.

I'd take out the words "almost" and "XML", and possibly also
"metadata", but otherwise I agree.

As you say, the fact that there is an XML format for topic maps is not
very important. I also think that thinking about topic maps (and RDF)
as metadata is misleading. They are really ontology languages and go
far beyond attaching metadata to information resources.

The whole point about them, and what separates them from the
hierarchical technologies we have been discussing here, is that where
both will say that "page X is relevant to Seoul" topic maps will say

  Seoul is a city,
  South Korea is a country,
  Seoul is the capital of South Korea, and
  page X is a tourist guide to South Korea
 
It's only the last line that is metadata. The first three, which are
the important ones for the purposes of this discussion, are not
metadata, as they are not speaking about information resources, but
about real-world concepts.

| Topicmaps define a very generic, powerful structure for certain
| types of metadata - the classification type. You define lots of
| topics, and then you define relationships between them and other
| funky things. Topicmaps are more complex than just a simple tree or
| a simple hierarchical faceted classification (although you can
| certainly express those as a topicmap) (read up on topicmaps to
| better understand the structure they provide).

This is true. (Well, "other funky things" is not very precise. :-)
 
| The question is: is using a topicmap structure to define the
| taxonomies of my site useful? Topicmap driven sites are often VERY
| cross referenced, very dynamic. You probably don't need that level
| of complexity. 

Well, some do, some don't. And even with topic maps you can choose how
ambitious you want to be with regards to the structure you create. You
don't have to spend 20 years developing the Ontology of Everything. In
many cases we've been able to create useful ontologies for clients in
a couple of days.

But you *do* need some degree of data modelling skills to be able to
create an ontology, and if you don't have that using the more
restricted technologies may be just as well. Whether it will be just
as well for the client is another matter, of course.

-- 
Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
GSM: +47 98 21 55 50                  <URL: http://www.garshol.priv.no >