Cart
[Home] [By Thread] [By Date] [Recent Entries]
KenNorth wrote:
> > Standard vocabularies for domains were pretty well
> > understood by scholastics in the middle ages. That
> > wasn't AI. It was common sense.
Common sense is a property of humans, not machines! The proper machine
encoding of common sense would be deserving of a Nobel Prize (in something)
IMHO. Perhaps that is why AI seems so easy but is in actuality so hard.
>
> Some time ago I wrote a Web Techniques article that discussed NLM's MeSH,
> domain vocabularies, and search engines such as WHIRL (which uses pattern
> recognition to support similarity searches). My suggestion was we should
> integrate the techniques. Use domain experts to build vocabularies, with
> machine analysis to continually scan and analyze the literature to surface
> new terms that have entered the vocabulary of practitioners.
>
A hot topic. For a preview of my RDF modelling of the healthcare spec which
gives a couple of RDF Schemas and their graphical representations (note,
this is but a preview and is not necessarily intended to be part of any
actual spec) see:
http://www.openhealth.org/ASTM/healthcare-model.gif
http://www.openhealth.org/ASTM/clinical-model.rdfs
http://www.openhealth.org/ASTM/E31.25-model.gif
http://www.openhealth.org/ASTM/E31.25-model.rdfs
and
http://www.openhealth.org/ASTM/operative-report.gif
http://www.openhealth.org/ASTM/operative-report.rdfs
the idea is that as actual operative reports, and other reports, are fed
into the system a semanticly meaningful database actually gets created.
Jonathan Borden
The Open Healthcare Group
http://www.openhealth.org
|
