# RE: Weights and Measures

From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
Date: Tue, 13 Mar 2001 18:25:52 -0000
To: "dehora" <bill@dehora.fsnet.co.uk>, <www-rdf-interest@w3.org>
Message-ID: <JAEBJCLMIFLKLOJGMELDIEEFCBAA.jjc@hplb.hpl.hp.com>
```
Comments in-line ....

> -----Original Message-----
> To: Jeremy Carroll; www-rdf-interest@w3.org
> Subject: RE: Weights and Measures
>
>
> Jeremy,
>
> I don't see that as an error of the M&S. To do what are you suggesting
> is to list all possible equivalent quantities (or as a many as
> you see fit).
> Normally, since here is only one or two that's fine. But why
> bother? Would we
> do the same for .../metres/centis/millis/...? Sure they have a family
> resemblance, but you still need an algorithm to convert between
> them. There
> isn't really any difference between 2 machines moving between
> pounds and kilos
> versus moving between kilos and tonnes. You just replace one algorithm for
> another.

No - perhaps I wasn't clear.
You are quite right to indicate that any solution to the units problem will
rely on algorithms somewhere to convert between various systems.

My dislike of the M&S graph for weights and measures is based on two issues:

1: Suppose I work in the metric system and download some imperial or mixed
metadata. Before using it I need to convert any weights and measures. If
these are expressed as in the M&S I need to *transform* the graph (i.e. with
edge deletion, replacement and addition operations). If these are expressed
in some other ways (such as the one in my previous message), then I can
simply add more edges to the graph; which is an easier and better behaved
operation. Moreover, if I am a fundamentalist believer in the metric system,
I can go round the semantic web, find other peoples imperial descriptions
and augment them (at least for people who upload my triples) with their
metric equivalents. This allows a third party to facilitate interoperability
between two incompatible agents.

2: The M&S graph suggests that the *weight* node is in units pounds; my view
is that it is the *value* 200 that is in units pounds; and hence the
description of <n:units rdf:resource="http://www.nist.gov/units/Pounds"/>
should be applied to maybe the triple <weightNode rdf:value 200>. My
previous suggestion of using n:Pounds as a subproperty of rdf:value also
works for me.

An argument in favour of this is that the sentence "John Smith's weight is
less than Mary Brown's weight" is independent of the units used. Also "John
Smith's weight is less than Franco Rossi's weight" should work even if
Franco Rossi checks in at 100 kilos (and doesn't know what a pound is).

>
> It doesn't seem incorrect to give only one measurement of John
> Smith's weight.
> If your machine doesn't know a particular unit of measurement, it can try
> to bind it to another and convert the amount. Or it can ask the source for
> any measurement mappings it might have. Or it can fail gracefully. Or some
> clever spark can try and siumlate the "Strategy" OO pattern in RDF whereby
> an RDF machine can give an input to another machine, telling it
> to convert to
> a given measure and return something it can manipulate.
>
> Bill de hOra
>

I like all of these suggestions ...

Jeremy
>
```
Received on Tuesday, 13 March 2001 13:25:57 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:44:29 UTC