Gerard, again:
>...The format can be applied via SGML using markup/tags. We need the same
collection of tags and with a standard meaning...
>The only thing is to think, think again, and make the standard collection
>of SGML tags to be used in medicine.
>
>Haven't I told you this before?
yes you have. *But* it is just that little matter of "make the standard
collection of SGML tags to be used in medicine" that seems to cause a few
problems. Those tags are what the non SGML people call their data model, or
Ontology. I know I have also said this before, (about a year ago) but it is
getting the world to agree on this sort of standard that is the problem, the
difficult bit - not how you represent it once you have the standards. The AI
community made the initial mistake of focussing on representation, to their
cost, but to their credit they realised that how you represent the knowledge
(e.g. the knowledge about a patient) is irrelevant, it is working out what
concepts, distinctions are needed that is important, and the relationships
between them. They worked this out about 15 years ago, surely we can learn
from them?
Once we have defined (if it is possible) these standard concepts and their
relationships, then we can represent, store and transmit them in myriad
ways, and I don't care to argue the merits of different systems - different
technologies are useful in different scenarios.
So can I make yet another plea that we drop discussion of representation as
irrelevant, and focus on the concepts and their semantics? This is the
interesting area.
Pete
---
Peter Johnson
[log in to unmask]
(+44) 1525 261432
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|