Re: [xml-dev] The triples datamodel -- was Re: [xml-dev] SemanticWeb permathread, iteration n+1

["Thomas B. Passin" <tpassin@comcast.net>]

Henrik Martensson wrote:

> I agree again. There are other factors though. For example, it is 
> necessary to factor in the consequences of not achieving conformance
> in terms of monetary loss, personal injury and death. When dealing
> with the automotive industry, military systems, or
> telecommunications, as I often do, misused tags are costly at best,
> at worst they kill people. Therefore, conformance has a very high
> priority.

Yes, it is true that the jobs I mentioned were not of that degree of
criticality.  The second one, the one where we went to parsing the text 
version rather than the html version, was much closer to being critical, 
though, because it deals with with informing hospoitals newly announced 
information about defects and recalls of medical supplies, equipment, 
and drugs.  There are potentially high costs for errors, both in human 
health and in legal liabilities.  Yet we so far have to accept 
hand-authored data that may vary in its format to some degree.  However, 
in this case we have humans in the loop, performing QC and editing 
functions.

Naturally you have to be more strict when the task is critical in its 
effects.  Still, I recall, back when I used to make actual hardware - 
not computer hardware,  computers were hardly used in the factory then 
if at all - our work required us to conform to some of the most 
demanding quality control specs, and to impose them on our suppliers. 
Nevertheless, as everyone who has been in a manufacturing environment 
will agree, every so often a batch of parts or other material will not 
meet specs.  Then there has to be a meeting of knowledgable people to 
decide what to do (the venerable Material Review Board).

And no matter how critical the mission, who can certify that a given 
piece of software is bug-free?  All you can do is to certify that it 
passed certain tests and metrics, and was developed in a certain way.

Of course, these considerations are no argument for not applying good QC 
or not worrying about bugs.  But even in the best situations, things go 
wrong or out of adjustment.  Having a schema out of sync with the actual 
documents is probably not uncommon, I would guess.

Cheers,

Tom P

-- 
Thomas B. Passin
Explorer's Guide to the Semantic Web (Manning Books)
http://www.manning.com/catalog/view.php?book=passin