At 14:46 -0600 2/13/06, Pat Hayes wrote:
The point I'm trying to make is this: The concept of "structuredness" is
relative and context-sensitive.
Hear, hear. Well said.
Pat Hayes
I second that. Yong revived an important point. In the SW October meeting
a year+ ago the importance of "context" was put on the short list of
important follow-ups. I haven't kept up .. but I am sure progress must have
been made. How RDF/OWL /ontology operate in different contexts and still
link to each other.
Many clients find that context helps bring a solution. One example where a
client receives 10-100's of spreadhseets on a daily basis. The key that
helped model a solution was using context to upload and join among the many
spreadsheets to diagnose the behavior of the patients through data gathered
from different scenrios. Context played a key role. Each spreadhseet was
captured during a context and the analysis was to extract factors that
affected the behaviour. In this case the complexity was in the variety of
contexts though the same construct (spreadsheet) came from and how to
intgerate together (semantic and data) to find the affecting features.
With the existance of many constructs like XML, HTML, RDF/OWL, relation
table (as in relational databases), excel sheet, text, etc a context is what
makes the ability to "glue" integarte all these constructs in syntax and
semantics together to extract information.
I am sure that more new constructs are being imagined on how to store and
capture information. Many of these constructs are not going away - like
data in relational databases!!! As I understood, one of the key benefits of
SW is how to "link" all together.
FWIW, Structured, unstructured and semi-structured, although non-precise
concepts in common language and (esp) philosophy, have well-defined and
precise meanings in database
jargon" -- most database books have decent definitions that are consistent
with:
unstructured - NL text
semi-structured - unstructured fields within a structured DB context
structured - relational model (or similar)
(those papers with technical definitions tend to get ugly and recourse to
relational calculus, so these overly simplified definitions should suffice
for now)
Well, OK, but those categories don't exhaust the possibilities. What about
NL text with RDF-based XML markup? What about XHTML with RDF markup? What
about common logic text? What about free text that has an associated
lexicon which is linked to concepts described by an OWL ontology?
At the integration consortium http://www.integrationconsortium.org/ some
intersting discussions revolve around that.
I view the definition of level of Unstructured vs. Structured makes it
easier to understand how well or easy is it to link or integrate one
construct to another construct. I guess that is why levels of
"structuredness" have been created within each organization to identify the
scale of complexity of integrating two constructs. I am not aware of any
standard around that yet? So one finds three levels in one group and
another organization sees them as seven levels!! I guess it is
"context-sensitive".
-Wafik