Re: ontology for units of measurement and/or physical quantities

On 28/09/06, Tim Berners-Lee <timbl@w3.org> wrote:
> While
>        box1  exterms:mass  [ex:value  2.4; exterms:units
> exunits:kilograms] .
>
>   might seem like a straightforward way, it turns out it is a trap!
>
> Presumably this one could write too
>
>        box1 exterms:mass  [ex:value  2.4; exterms:units
> exunits:kilograms] ;
>            exterms:mass  [ex:value  2400.0; exterms:units exunits:grams] .
>
> You would expect mass to be a functional property, to the two Bnodes
> must
> be owl:sameAs each other, so we can conclude that the masses are the
> same:
>
>        box1 exterms:mass  [
>                         ex:value  2.4;
>                         exterms:units exunits:kilograms;
>                         ex:value  2400.0;
>                         exterms:units exunits:grams]
>
> bzzzt! tilt! that means we have a mass with two values and two units.

Nice catch. exterms:mass in the original example would maybe have been
better named exterms:massMeasure, to avoid the (natural) expectation
of it being a functional property.

> A much *better way* is to use the unit as a connecting predicate.
>
>        box1  exterms:mass   [ exunits:kilograms 2.4 ].
>
> You can then write quite validly
>
>        box1  exterms:mass   [ exunits:grams 2400.0;
> exunits:kilograms 2.4   ].
>
> This takes less space and can be easily manipulated, and doesn't
> throw up
> horrible errors.
>
> A unit kg is the relationship between 3kg and 3.

Yup, that looks a lot better.

[snip]

> PPS: can't you even say, using units predicates as multiplication or
> division of scalar quantities
>
> @prefix : <....exunits#>.
> @keywords.
>
>    box1  height 3 ^ m ;
>         speed   3 ^ m!s ;
>         acceleration 9.81 ^m!s!s .

Not sure I grok what you've got there, is ! a predicate (on the
literal) as "per"? (Is this all N3?)

In the defence of, er, the devil, I'd have to note that the source of
the example, the RDF Primer[1], doesn't cover the layer which includes
functional properties (is the approach a reasonable
model/approximation? Ok,  your point above says not). Can't help but
notice too that the doc "has been endorsed by the Director as a W3C
Recommendation"...

(does the n-ary relationships note [2]  cover this I wonder?)

Cheers,
Danny.

> > [1] http://www.w3.org/TR/swbp-n-aryRelations/
> > [2] http://www.w3.org/TR/rdf-primer/#rdfvalue


-- 

http://dannyayers.com

Received on Thursday, 28 September 2006 22:38:29 UTC