[Sigia-l] Eliminating categories in favour of tagging
Alexander Johannesen
alexander.johannesen at gmail.com
Sun Mar 12 17:17:59 EST 2006
On 3/11/06, dbedford at worldbank.org <dbedford at worldbank.org> wrote:
> Just a clarification - I think this will help you to better understand the
> underlying 'architectures' which is what is supposed to be the focus of this
> list.
"You" the members of this list, or "you" Alex and Andrew?
> When you frame the argument in terms of free form or controlled form, then I
> think the issues become clearer. In a nutshell, the issue with 'entirely free
> form' terms is that you create a mess that has to be cleaned up later by
> somebody -- and rarely is ever cleaned up.
What I argued for was a 40% controlled and 60% uncontrolled top-down
structure rolled into one. I've used, implemented and played with many
variations on this, but usually the 40/60 rule seems to be king.
> There is a happy medium between
> free form and controlled -- which pertains to semantic analysis.
Depending on who you imply does the analysis, I'd agree if you're
talking about the human factor at the run-time level. If you mean
semantic inference (pre-processed or real-time), I'm a little more
sceptic; not because it can't be done, but because it is expensive and
quite often not worth it. (There's a reason social engineering is so
popular; dump all the hard [as in difficult] work on your unsuspecting
userbase)
> I think it would be great if this list could begin a discussion of these issues
> - the free form or controlled discussion is a few decades old. We have moved
> beyond it.
I think that statement is a bit broad and sweeping; even if something
is old doesn't make it unrelevant or solved. As far as I know,
politics were still explored thousands of years ago, and we still
haven't solved that one. :) (Same goes for list behaviour and ethics,
but I digress ...)
> Are you all aware of the Semantic Technologies conference in
> progress this week in San Jose? You might want to follow what is happening in
> that community.
That community. Heh. Some of us have been part of "that" for years, me
since 1989. The thing is, and it has been the thing since AI started
in the 60's, it still can't outperform a person with some papers, a
cup of coffee, and the thought "hmm what's going on here?" Which, in
essense, is what IA and the focus of this list is to me. Semweb stuff
is a cute tool, though.
Alex
--
"Ultimately, all things are known because you want to believe you know."
- Frank Herbert
__ http://shelter.nu/ __________________________________________________
More information about the Sigia-l
mailing list