On 17/08/2006, at 11:28 PM, Ian Cheong wrote:
At 10:43 pm +1000 17/8/06, kuang oon wrote:
On 17/08/2006, at 8:49 PM, Ian Cheong wrote:
[...]
It seems likely to me that computers will conquer natural
language processing faster than clinicians en masse can navigate
a controlled vocabulary with accuracy and speed.
That may turn out to be true. Natural language is equivalent to
that flat plate supported on the back of a giant <http://
en.wikipedia.org/wiki/Tortoise>tortoise. that holds up the universe.
But, "What is the tortoise standing on?"
"But it's <http://en.wikipedia.org/wiki/Turtle>turtles all the
way down."
We still need the turtles....meaning some distinct layers of
symbolic codes for representation of data.
The point is we need a team of turtles, a linguistic stack, to
support that flat plate that supports the world :)
Interesting...
Medical natural language is built on a stack of symbols and rules
for symbolic combinations:
* medical grammar in common usage, based on english grammar
* medical terminology described in medical dictionaries
* English grammar
* English and adopted words made of letter and as described in
dictionaries
* roman alphabet (symbols)
In some respects, an artificial medical language is a duplication
of that stack with another, which tries to mean the same but
doesn't quite exactly.
A key aspect of language is that it is uncontrolled. It happens and
it is generally documented after the fact.
So will medical language autocracy rule over democracy?
Hi Ian,
That large flat plate that supports the world is hard to balance. To
make it happen, there are at least 2 schools of thought about how
this plate should be supported:
1) a single large tortoise with a magnitude of 2**64 .
Or
2) a specialized team of turtles that stacks vertically like a
building scaffold. The team members have weird names like
'docletalk' 'sheep' 'doclescript' and 'docle' :-(
Kuang
Team turtles
"The fear of turning turtles melts away when you think of the turtle
soup."
_______________________________________________
Gpcg_talk mailing list
[email protected]
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk
