[Sigia-l] Disposable navigation systems

[Alexander Johannesen]

Hi,

Natasja Paulssen <natasja.paulssen at ordina.nl> wrote:
> from an informational viewpoint metadata and navigation structure
> should be independent of each other. I believe that talking about disposable
> navigation systems will help people understand the importance of this
> independency better.

So in a sense you can plug different navigation interfaces on top of
the same sets of data, and get various 'meanings' out of it? Sure,
look to any application out there that tries to enforce a clear
separation of interface and data, be it MVC applications to semantic
HTML with CSS on top to ... uh, quite a number of them.

It is, in my view, more a practical hurdle than a philosophical one.
People understand that the vastness of data implies semantics even
though the semantics are not structured in our applications. People
very often *want* structured and cumbersome navigation, mostly because
they're used to it (traditions, culture, etc), and not nescessarily
anything else, even if they might be better.

Working with Topic Maps, which has complex structures to handle
complex semantic data,  has taught me that even if you *can* create
better means of navigation (and by better here I'm talking about the
path from start to finish in resource discovery) they most often
*prefer* more cumbersome traditional tree-structural means. (I've got
a rant coming on about why classification taxonomies are so sexy yet
fail to please, but I'll leave it for now ;)

> Faceted navigation may certainly leed to a disposable navigation system.
> Yet designing with the temporary feel of disposable in mind might leed
> to new ideas, I hope.

Yes and no; yes, because people get stuck in a viewframe and don't
catch on to context not expressed within, but also no, because for
knowledge to stick we need more specific context. The more fuzzy the
context, the less we are likely to understand them.

This is why ontology work is so in the wind these days; the pipedream
goes that once you've defined your data in your ontology, you can
tweak the ontology (or create separate ontologies to the same dataset)
without losing semantics, but in every practical and philosophical
term this has been proven to be misleading, or even false. In a sense,
this is why folksonomies almost works, taxonomies almost work,
classification schemes almost work, but none of them really work by
themselves.

To get a truly detached navigation of a dataset you can say very
little about your data, I'm afraid.

> The ultimate example being: feed the outline of
> your document into a search engine that generates an information hierarchy
> on the spot. Interesting e.g. for checking business cases to see what other
> projects claim the same benefits...
> Does this help?

Well, if you just want to play around with an automatic ontology tool
(which the above sounds to me) then sure, you can fix your mistakes
and create better semantics ... but the thing is, do you always *want*
better semantics?

Think of the old problem with chucking keywords into your HTML page
for search-engines to spider; to few, and you didn't get listed. Too
many, and you got hits for almost any stupid search. At some point
inbetween too many and too few lies the (assumed) correct answer.

When you feed your outline of your document into a search engine that
generates an  information hierarchy on the spot, is the result right
or wrong? My guess is that it would provide a hint if you had a
piss-poor document fed, but in the end, it is just a visual aid,
right? What would this tool bring you?


Alex
-- 
"Ultimately, all things are known because you want to believe you know."
                                                         - Frank Herbert
__ http://shelter.nu/ __________________________________________________