Re: ANNOUNCE: Medical Words v0.1

[Elizabeth Barham <lizzy@soggytrousers.net>]

Andreas writes:

Gutan tag,

> Thanks for the announcement.  We'll see how we can turn this into a
> usable Debian package.  Do I understand it right that its primary
> usage might be spell checking or do you see any further
> applications.

   This release is simply a listing of words so, yes, its main use is
spell checking.

   There is another part, however, that has not been released and I'm
currently a bit squeamish about releasing it and I need to do various
preliminaries prior to doing so. This part consists of word parts,
most of which are Latin, and their definitions. It contains things
like:

hepat,o,liver
iatr,o,treatment
leuk,o,white
nephr,o,kidney
neur,o,nerve
onc,o,tumor
ophthalm,o,eye
oste,o,bone

   This could be used in some kind of grammer checker or a "medical
word help" sub-application/process within an editor, for example, or
perhaps an "automatic word definition" thing. There are also prefixes
and suffixes and by combining these a definition is made:

radi,o,x-rays # root
logy,study of # suffix

radiology, x-rays study of (study of x-rays)

   The reason I'm a bit hesitant about releasing these involves
copyright issues as this is not the same as a listing of words which I
know cannot be copyrighted; definitions *can* be copyrighted but these
definitions are so generic I question whether they can be copyrighted
as well. So I'd like to verify this situation prior to releasing it.

> PS: What about your usfda-ndc-drug-info project.  If I remember
> right we stalled at the fact that I was unable to compile the
> package.

   Its still there and that problem seems more akin to a
compiler/library issue and not with the specific package; when the
various libraries for the next release are standardized this problem
should go away.

   However, I do question the use of the database data itself outside
of the medfo project but it seems a shame that the data could not be
used by other projects. I mean it can be done but I wonder if this is
the better approach. What worries me is that the FDA data format may
change so much that it will break the applications currently using the
data although an easy work around would be to version-ize the
databases *format* as well as perhaps a date versioning.

   For example, if the data follows the format currently in use, it
could have a "-1" after the name, followed by the date of the data's
release:

           usfda-ndc-drug-info-1_0.1-20030331
           usfda-ndc-drug-info-1_0.1-20030730

The different dates corresponding with the date the FDA released the
tables. If, however, they change their format to such an extent that
current applications working with it will break, its sub version could
be increased:
         
           usfda-ndc-drug-info-2_0.1-20040331

For the deb packaging, medfo, for instance, would depend on
"usfda-ndc-drug-info-1" and incremental updates would take care of any
new releases of the same format. Any new format - new table names, a
new column, etc - would then become part of the "-2" release.

Elizabeth