On Mon, 29 Nov 2010 17:34:14 -0500, Denenberg, Ray wrote
> (second of two messages)
> These all have distinct meanings.
> It means just that, "unknown". So for 199u, the 'u' may be replaced
> by any single digit. [Note: we have not settled on 'u' as the
> "unknown" character. We just haven't found a better one yet.]
> There has been discussion about whether and how (1) precision might
> be assigned to "approximate", and (2) probability might be assigned
No. My suggestion of explicit precision was not to address approximate but
the "digit unknown" case. The idea was to ***replace*** the u masking model
with an explicit specification of precision in many use cases.
Again if you look at 199u (where the 'u' may be replaced by any single digit):
what is it essentially saying? A date in the 1990s, viz. a date with
This model does not cover, of course, all the cases as some oddball maskings
such as 1uu1 don't have an analog in precision but it covers, I'd suggest,
all the typical use cases.
Precision we could break down into "readability" and "repeatability". By
1990s we've implicitly aligned its "readability" (the decade) with its
repeatability. We might want to separate these two concepts to allow
predicates to express their features independently.
The predicate "approximate" is more something like tolerance.
> to "questionable. I hope we agree that qualification is not
> necessary for "unknown" and "uncertain".
The predicate "questionable" is about "accuracy". I've not suggested we
get too deep but keep to the simple predicate whose semantics would be
"Possibly correct" (might be but not likely) or less.
Of the simple model
- a) Known to be correct (observed, documented etc.)
- b) Likely correct ( p> 50%)
- c) Possibly correct (Might be but not likely)
- d) Likely incorrect (The date is expected to be wrong p ~ 0)
- e) Unknown (certainty unknown).
it would put "likely correct" as known to be correct. This, I think, is
effectively how people would assign ? to dates. If they think "its probably
right" they won't generally bother with flagging something as "questionable".
Its only when they strongly suspect that it could be wrong or even likely
incorrect would they speak up...
> If indeed there is a need to assign precision and probability to
> approximate and questionable then I propose the following basic
I've suggested that since dates, like other measures, have qualities of
accuracy, readability and repeatability, we may want to explicitly suggest
them. They are, after all, being implicitly expressed all the time.
Edward C. Zimmermann, NONMONOTONIC LAB
Basis Systeme netzwerk, Munich Ges. des buergerl. Rechts
Office Leo (R&D):
Leopoldstrasse 53-55, D-80802 Munich,
Federal Republic of Germany