[Sigia-l] Eliminating categories in favour of tagging

[Alexander Johannesen]

On 3/13/06, Seth Earley <seth at earley.com> wrote:
> Tagging is not "the new best practice in metadata generation for blogs".
> Uncontrolled vocabularies are a giant leap backward into the past. (Why
> would tags be a 'poor man's uncontrolled vocabulary'?  An uncontrolled
> vocabulary is an uncontrolled vocabulary.  It means that the person applying
> metadata is deciding on terms on the fly and can choose whatever term they
> think is appropriate.)

"Tagging" as a term has, as others have pointed out, been around for a
long time. In my world of libraries and cataloguers it is known as
free-tagging and gets stored in the wonderfulness of MARC records, and
is a rather expensive and holy ritualistic approach. "Tagging" of
today is poor-mans choice because the IT sector has made it easy and
affordable. But this easiness comes at a price; the lack of quality.
Where as in a library free-tagging setting there are control-factors
for things like spelin and over/under use of terms, in a world of
Flicr/del.icio.us these lack of controls cause what's known as tag
fungi. :)

Of course, brute force remedies can fix someof these (auto
spell-checking being the somewhat simpler).

> You mention faceted classification, but I don't think faceted
> classifications can be uncontrolled.  Terms in a faceted classification need
> to be orthogonal, that is mutually exclusive, there can be no overlap.  (I
> suppose one could argue that terms could repeat from facet to facet but be
> in different contexts in each facet but I don't think this would be a good
> idea.)

I could argue that, of course. But when I build facetted systems, a
facet is nothing more than a property of that item (in technical
terms; there is a binary association in any one-way relationship which
can be inferred), and you can group and structure all binary
associations in pretty much any way you like it. You can have a facet
which is unique to an item; that's just presentation. Of course it
benefits the system that facets are somewhat evenly distributed across
your dataset (especially if you want it all to look nice :), but there
is nothing preventing me for having a facetted system of more unique
than general facets.

> So I would suggest one not throw out controlled classifications but augment
> them with collaborative tagging and then review collaborative tags for
> inclusion in controlled classifications.

Hmm, I never suggested anyone throw out controlled classifications for
tagging either, so I guess we're in agreement. :)


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