Re: Comment to ONC recommending RDF to help "standardize the standards"

Hi Grahame,

On 05/01/2015 06:05 PM, Grahame Grieve wrote:
> HI David
>
> I cannot let this go by without some comment.
>
>     Therefore implementers are required to create their own workarounds
>     to bridge these standards and overcome the incompatibilities.
>     However, because implementers use different approaches, data still
>     lacks interoperability between implementers.
>
> you make it sound as though the standards are the problem, rather than a
> symptom of the problem.

Oops.  We certainly did not mean to denigrate standards!

> Even if all the standards completely agreed,
> implementers would still have to do this. And it will be a long time
> before we can get consistent standards across the whole stack - some
> levels we haven't even started to have discussions about

Agreed.  But I also think that RDF can help in the evolution toward 
consistent standards.

>
>     The ONC should add a recommendation for a universal information
>     representation that can accurately capture the meaning of any
>     healthcare information, spanning all ONC-mandated standards,
>     regardless of the the data formats, data models,  or vocabularies
>     prescribed by those standards.
>
> we've already had one of these, and we know what the outcome is - the
> universal representation steadily becomes it's own goal, more important
> than the outcomes that are sought.

I think it depends on how it is done.  I think a more bottom-up, 
collaborative approach can succeed where previous approaches have not. 
RDF has some key benefits that help in this regard: being independent of 
data format; cleanly accommodating distributed extensibility; and 
supporting inference.

>
>     Suggestion 2: Recommend RDF as the best available universal
>     information representation
>
> This completely misses the point; for the kind of disagreement that your
> comments seek to address, RDF is just a format that has few semantics
> (and the ones it has work as much against the goals as for them). I'm
> not against getting common RDF representations for standards
> (obviously), but it's just moving the deck chairs around: it won't make
> any difference at any level that matters.

I agree that it does not solve the problem in and of itself.  There will 
still be semantic differences between similar concepts in different 
standards, even if the information is all expressed in RDF.  But I think 
the adoption of a common information representation is an important 
first step, because: (a) it allows those differences to be exposed and 
more visible in that common representation; and (b) it allows semantic 
relationship between those concepts to be expressed in that same common 
representation.

David Booth

Received on Saturday, 2 May 2015 00:05:11 UTC