I'd say there are at least three ways of expressing imprecision:
a) use a range (A.D. 320 - 350)
b) use a larger unit (4th cent. CE)
c) use a modifier denoting imprecision, such as "circa," "around,"
"about" (c. A.D. 260)
The same issues apply to any kind of measurement also. The question
is whether any of these are properly expressed in markup rather than
as part of the measurement. In Epidoc, the answer is pretty clearly
yes. We don't feel comfortable saying:
<gap reason="lost" extent="c.12 characters"/>
we'd rather say (as we do):
<gap reason="lost" extent="12" unit="character" precision="circa"/>
In a) and b) above, the imprecision is expressed pretty explicitly in
the date itself. Likewise in the Sciences you get quite precise
measures of imprecision with margins of error, the right number of
significant digits and so on. But it's common practice in the
Humanities to say "this number is fuzzy" and leave it at that without
quantifying the fuzziness in any way. I'll happily concede that you
might not always want to signal that fuzziness as part of the markup,
but I do think it's sometimes very useful to be able to do just that.
* Hugh A. Cayless, Ph.D.
* Head, Digital Library Technology R & D
* UNC Chapel Hill
* [log in to unmask]
On May 19, 2007, at 1:03 PM, Gabriel BODARD wrote:
> Tim Finney a icrit :
> > I second Gabriel's call. It seems to me that both accuracy and
> > precision must be specified when describing uncertainty.
> This isn't actually what I was proposing, since I have gone on
> record before arguing that in almost all cases of mark-up we can
> use less specificity in expressing certainty and uncertainty. In
> the markup equivalent of an editor putting a question-mark before a
> restoration or interpretation, all we really *need* to be able to
> do in markup is say cert='high' (default = no question-mark) or
> cert='low'--and maybe have a target pointing to what aspect of the
> restoration is flagged as uncertain.
> What I am suggesting is that we need a *different* mechanism to
> indicate imprecision of numerical data; again all an editor
> normally does is either to write "12 cm" (precision='high') or "c.
> 12 cm" (precision='low'). It's actually rather unlikely that we'd
> want to use both precision and certainty at the same time, unless
> we're actually marking up statistical estimates (in which case
> there must be a technical namespace we could use to enhance the TEI
> Re Syd's objection to the use of @precision with date, I think I
> disagree. Saying "A.D. 240 - 260" is not the same thing as saying
> "c. A.D. 250" (editors do both, and presumably mean different
> things by them). Presumably your argument is that <date
> notBefore='0250' notAfter='0250' exact='none'> would be equivalent
> to <date value='0250' precision='low'>?
> Leaving aside dates, however (which until Syd's sneak preview
> appeared to already allow my proposed @precision), there is no
> similar objection to the use of @precision to mark the 'circa'
> status of gap[@extent], space[@extent], or measure[@value], is there?
> Dr Gabriel BODARD
> (Epigrapher & Digital Classicist)
> Centre for Computing in the Humanities
> King's College London
> Kay House
> 7, Arundel Street
> London WC2R 3DX
> Email: [log in to unmask]
> Tel: +44 (0)20 7848 1388
> Fax: +44 (0)20 7848 2980