From: Samantha Atkins (email@example.com)
Date: Mon Jan 07 2002 - 03:05:47 MST
There is no reason the structure added to unstructured data has
to be brittle. Some kind of semantic net of information
extracted for instance need not be all that brittle at all.
Also, the amount of structuring that is actually needed to make
unstructured data significantly more useful doesn't necessarily
require that much expertise or expensive processing. A relative
handful of low-level hacks spread across all pages of certain
types would already be a huge improvement.
Jeff Bone wrote:
> Ben Goertzel wrote:
>>Well, I tried to explain this, but apparently I was not clear enough.
> My apologies, I've apparently been trying to make my point a bit too subtly.
> IMO, the process you describe --- using *any* tools to turn unstructured data
> into structured data --- is dangerous and counterproductive. Ontologies are
> inherently brittle, static things, and require lots of care and feeding. Using
> expensive tools to turn unstructure into structure *today* just leads to more
> noise and cost tomorrow. I don't dispute that this can and will happen ---
> indeed, *is* happening. My point is it isn't a good thing, even temporarily.
> Consider: over 90% of the lifetime cost of storing data isn't in the storage
> media / hardware, it's in the operational and support --- i.e., information
> management --- costs. And the interesting (and counterintuitive) thing is that
> structured data has *much* higher lifetime TCO than unstructured data.
> Unfortunately, we're going to spend a whole lot of effort to create additional
> (and brittle) structured (and metastructured) information that will ultimately
> be "noise." I wish all that effort --- and all the "cheaper, dumber tools"
> effort to extract value from that dubious process --- was instead spent on other
> more useful things like:
> Good, useful tools for dealing with unstructured data needn't be costly. :-)
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:37 MDT