From: Jeff Bone (jbone@jump.net)
Date: Tue Jan 08 2002 - 00:01:14 MST
Samantha Atkins wrote:
> There is no reason the structure added to unstructured data has
> to be brittle.
Well, I'm just not sure that's true on a fundamental level.
> Some kind of semantic net of information
> extracted for instance need not be all that brittle at all.
Okay, this I *know* not to be true. Brittleness in this case is an unavoidable
artifact of the mutability of objects (files, etc.) in today's information storage
paradigm. Because an object from which semantic information is "extracted" may later
be changed, having a separate semantic net (metadata of any kind, really) overlaid on
top is an *inherently* brittle arrangement. Unless the storage substrate itself
becomes smarter / can understand the semantics of what is stored, you're always going
to have a potential for metadata and data to be out of sync. And IMO, that's worse
than not having the metadata at all.
Further, ontologies are inherently brittle because they are models of the world, and
these models evolve over time as our context and our understanding of the world itself
evolves. Any semantic network which seeks to capture such an ontology *is* doomed to
obsolesence. (To illustrate this: how many times in your life have you rearranged
your own home directory filesystem to better organize your stored bits?)
> 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.
I'm completely with you regarding "annotation"-like external markup and addition of
dimensions to otherwise-unstructured content. But IMO, the grand vision of the
semantic web goes much further than that, into a relatively dangerous territory. Do
some reading about the history of the hierarchical database paradigm for a starter /
refresher course in the danger of reference-based structures as a storage metaphor...
Finally, I'm not convinced that structure isn't *inherently* harmful. IMO, it's a
poor man's workaround for lack of intelligence in general. There is no evidence to
suggest that information is stored / accessed / used in a structured way in biological
intelligence. I'm not sure why we think it's so important in artificial information
processing systems.
jb
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:37 MDT