How many times have I said that *standardizing on the number of significance does not say anything about the the level of uncertainty.*
Again, a rough approximation of the uncertainty is obtained by the *radius*.
Why do you assume that the measure of extent or accuracy follows a normal distribution?
There is also *no* assumption in *geo* or in the ietf.org proposal that these are typed as a float. They are a string of characters.
The standardization on the number of significant digits is to standardise the resulting geo *urn* as a string.
According to the ietf proposal (if it is adopted) is that "geo:41.53000000,-70.67000000" and geo:41.53,-70.67" identify the same resource, but they are not the same urn and will not be seen as the same urn in a triple/quadstore.
This means that you will not be able to browse between occurrences associated with the same GPS reading, or search for them via Google.
"geo:41.53000000,-70.67000000" and "geo:41.53000000,-70.67000000" will be interpreted as the same urn by triplestores, the same string by Google and the same location by software that correctly interprets the ietf.orgstandard.
Rather than having all the providers rounding to different numbers of significant digits, standardizing on them makes the resulting data more comparable.
- Pete
On Tue, Mar 1, 2011 at 7:40 PM, Paul Murray pmurray@anbg.gov.au wrote:
On 27/02/2011, at 6:01 AM, Bob Morris wrote:
"Note: The number of digits of the values in <coordinates> MUST NOT beinterpreted as an indication to the level of uncertainty." The section following is also interesting, albeit irrelevant for your procedure. It implies that when uncertainty is omitted (and therefore unknown), then "geo:41.53000000,-70.67000000" and "geo:41.53,-70.67" identify the same geo resource.
This follows from the definition of xs:float, which RDF (and presumably geo) borrow. These two strings are each representations of the same IEEE floating-point value. Something with a precision is a different datatype.
It seems to me that georeferencing needs a notion of density - some way to express a location as a value at a point and a standard deviation (we assume a normal distribution). A rectangle becomes an integral of these densities at each point - standard deviation of zero corresponding to the usual sharply-defined rectangle. You'd probably also want to supply a cutoff value.
If you have received this transmission in error please notify us immediately by return e-mail and delete all copies. If this e-mail or any attachments have been sent to you in error, that error does not constitute waiver of any confidentiality, privilege or copyright in respect of information in the e-mail or attachments. Please consider the environment before printing this email.