Alvin Marcelo wrote:
>
> Again, dear researchers, you have to reach out to the practitioners in the
> list because they are the ones who will give life to your models.
[snip]
> I am sure many of the
> other practitioners in the list speak different languages, but I speak SQL
> and that's is where my bias is and that is why I insist on it in previous
> posts.
>
Ok. Let me try again by quoting from the GEHR design document
(http://www.gehr.org/architecture.pdf) and see if any of this makes
sense to the practitioners on the list:
> The need for a model is a consequence of the need to make explicit the semantics of
> the record and its operations. Without a model whose primary purpose is to express
> semantics, implementation expressions such as protocols or database schemas
> whose primary job is to concretely define tables or data packets end up also trying
> to implicitly express semantic concepts, usually imperfectly.
>
This statement says the SQL data tables are not sufficient to represent
the model.
>
> Thus the problem of how to get from an abstract model to an actual health record is
> addressed by providing expressions of the model in such formats. As mentioned in
> the Requirements, there are two ways to do this: via a binding (set of rules via
> which a transformation can be made at any time), or via a mapping (a complete
> expression of the model in the target format).
>
The formats referred to are:
>
> For the typical modes of communication -file,
> database, object distribution or messaging - there are various standard and not-so-
> standard formats.
>
And this would include such things as ODMG ODL, SQL, OMG IDL, HL7 and
XML.
finally, in figure 2 of the architecture document an application is
schematically represented as the following:
Application <uses kernel api> kernel <uses persistence api> DBMS
So in this model, the application programmer no longer 'sees' the DBMS
interface, that is done by the kernel. The application programmer
'sees' the kernel api, which is an object oriented collection of
interfaces. The clinically relevant semantic information is represented
by the kernel interfaces. The database mapping is a system design issue
between the kernel and whatever database technologies are employed; flat
files, indexed files, RDBMS or OODBMS and the application programmer has
no need to become involved.
That means that the core concept of a kernel in the GEHR sense
provides a layer of object abstraction. I imagine this is not quite the
same sense in which Alvin has been using the word kernel. If I am right,
I hope to have identified a source of mis-communication and perhaps
clarified it somewhat.
S/MIME Cryptographic Signature
