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

Hi Pierre-Antoine,

We use n-ary relations (we call them 'templates') such as:

    <oim:ST-VESSEL-3401-002 rdf:ID="ME03_ST-267634">
        <rdfs:label>vessel V-6060 has a design pressure of 100
        <part7:possessor rdf:resource="#ME03_347621-20060306T1357Z"/>
        <part7:numericalValue rdf:resource="#XSFL_100"/>

    <part4:XmlSchemaFloat rdf:ID="XSFL_100">

For each template class we define, in OWL, the Restrictions on its
The 'possessor' is a "temporal part" of ME03_347621 (that temporal part has
been declared separately).
'DesignPressure' and 'Barg' are defined in our Reference Data vocabulary, as
per ISO 15926-4.
The numericalValue is an instance of XmlSchemaFloat, with a 'content'
property expressed in the XML Schema float datatype.

In forementioned Reference Data vocabulary we also have defined the
relationship between Pressure and the Scale, and between that Scale and its
NumberSpace (here: -1 to +infinity). So our software can validate whether or
not the value of 100 falls in that instance of NumberSpace.

You, and other readers and contributors of this forum may think this is too

But this is only a simple problem :-) Try to represent a specification for a
centrifugal compressor or the like, in such a way that the computer system
of a supplier at the other end of the world, who doesn't know you, can
extract exactly what you mean. And 20 years later some engineering outfit,
handling a revamp of the plant, still can understand what you meant.

Text and pictures are a lot simpler (for humans), we still can understand
what Shakespeare wrote or Rembrandt painted.The main problem is that today's
computer systems are still rather unintelligent (as compared to the
capabilities of the human brains).

The problem with data is that they are almost always implicit, i.e. we leave
out a lot of information because we assume that the (human) recipient can
fill in the relationships between the data based on the context (e.g. it is
on the same document) and on domain expertise. OWLites will say that we have
reasoners for that, but I don't believe that any manager will be willing to
rely on their outcome. So we (the process industries) decided, almost 20
years ago, to make our exchange information as explicit as is reasonably
possible in order to make it more understandable for computers, and to allow
for integration of lifetime information about an entire plant, its
components, and its streams. All of that in a neutral format that is

The Semantic Web, with its RDF and OWL (Full!), SPARQL, and SOAP, has come
in our world at the right time. We love it! But if N years from now the IT
industry has totally forgotten these technologies (remember COBOL?) our
lifetime data can easily be migrated to whatever gizmo's are then the utter
wisdom. Because we rigorously modelled them.

I'll get off that soapbox :-)


PS to ISO 15926 insiders: the above example has been amended somewhat for
editorial reasons.
Hans Teijgeler
ISO 15926 specialist
+31-72-509 2005

-----Original Message-----
From: [] On
Behalf Of Pierre-Antoine Champin
Sent: Thursday, October 12, 2006 11:27
To: Joshua Tauberer
Subject: Re: ontology for units of measurement and/or physical quantities

Joshua Tauberer a écrit :
> The big drawback is that dealing with these literals requires a 
> special parser, but it saves creating a new bnode for each measure (i.e.
> mysphere mass [ grams 1000 ] ) and I think is a little easier to 
> understand compared to having a bnode that represents the notion of 
> the mass itself independent of its measure.

There is indeed a tradeoff to find between encoding things in "complex"
literals and making them explicit in resources.

Note that checking the consistency of measurements as blank nodes requires
ad-hoc implementation as well : it is not possible to state that any
resource with :grams "1000" must have :kg "1"... An the problem of rounding


No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.0.408 / Virus Database: 268.13.2/472 - Release Date: 11-Oct-06

No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.0.408 / Virus Database: 268.13.2/472 - Release Date: 11-Oct-06

Received on Friday, 13 October 2006 05:38:12 UTC