[Om3] UnitsML

[Professor James Davenport]

I was kindly invited to the Telco this morning. The following 
points came up:
1) NIST pointed out that OMFOREIGN existed, so (JHD's interpretation)
   <OMA>
     <OMS name="eq" cd="relation1"/>
     <OMS name="metre" cd="units_metric1"/>
     <OMFOREIGN> UnitsML </OMFOREIGN>
   </OMA>
   would be legal as, say, an FMP.
2) NIST pointed out that dimension is necessary, but not sufficient, as in 
   energy versus torque, both of which are force times length. Quantities 
   are (currently) not as formal as dimension.
                                                     JHD to discuss w/NIST
3) UnitsML currently specifies a conversion as text, a service (via WSDL), 
   or a linear equation. JHD pointed out that these equations should be 
   formal, i.e. in MathML-C/OM.
                                                     JHD to discuss w/NIST
4) JHD noted that, because of OMFOREIGN in one direction and 'external' in 
   UnitsML in the other, either can be included in the other, so it would 
   be desirable to have fixed inclusions with known properties of 
   roundtripping (preferably identity!).
5) NIST noted that UnitsML has "reported digits", so 9.17 \ne 9.170.
   JHD said this wasn't directly possible in MathML as far as he knew.
   This is part of the general issue of treating uncertainty.
   JHD wonders privately whether number+interval doesn't help.
                                                           Any comments?
6) UnitsML is working on symbol descriptions: JHD said that, as MathML 
   includes Unicode, this could be done via MathML(-P).
7) UnitsML has the concept of "counted item", allowing, say, 
   electrons/sec.

Most of the real information is not accessible at OASIS, but at 
http://unitsml.nist.gov, in particular
http://unitsml.nist.gov/Schema/Documentation-0.9.17

James