unitDictionay additions made should we do a supplemental release?

[Matt Jones]

Scott,

> BTW, this excercise makes me wonder if we aren't re-inventing the
> wheel, OR inventing a wheel that we shouldn't have to.   Has anybody
> reviewed:
> 
> http://www.unc.edu/~rowlett/units/index.html
> 
> He certainly seems to have some expertise in the area.
> 
> And, what about NIST itself, or some other government standards body?
> Why are WE having to do this and keep it correct and up to date?  I
> definitely believe the unit dictionary should be de-coupled from EML
> in the next release.

I don't think we are completely reinventing the wheel :).  There are 
certainly good sources of information about many units, such as the web 
site you list, and NIST itself.  What these don't seem to do is provide 
a *formal and quantitative* framework for defining the units in their 
lists and the relationships and combinations of units that might be used 
in a particular application.  That's what STMML provides.  I'm sure the 
CML folks have a dictionary built in STMML that covers much the same 
ground as us.  We should ask them for it, and see about collaborating on 
  a common set. I might actually have part of it in an email that Peter 
Murray-Rust sent to me.

When writing the SEEK proposal, I also looked for things like this.  The 
only thing I came upon was a quantitative ontology of engineering units 
at the Stanford KSL site (http://www-ksl.stanford.edu).  The theory (and 
an explanation of the approach) for this ontology is in the Gruber and 
Olsen paper titled:
    An Ontology for Engineering Mathematics
    http://www-ksl.stanford.edu/knowledge-sharing/papers/engmath.html

Its a good start, but it only defines a few units (60).  The ontology is 
written in lisp, and it depends heavily on other ontologies that are 
based on the KIF framework.  Its hard to get to these ontologies out, 
but you can get to them from the Chimera ontology browser/editor on the 
KSL site.  Their approach uses the term 'dimension' in the way that 
STMML uses 'unitType' if I read it correctly.  Its pretty tough to read 
a lot of lisp though, so I haven't delved too deeply into it.

There are probably other works out there like this that I haven't found. 
  So, I think the answer to your question about reinventing wheels is 
yes and no, but what we are doing is more practical and less 
theoretically based in ontologies.  I think the approach they tak would 
allow for the semantics issues I raised in my last email much more that 
STMML does (but that's just speculation on my part).

Matt

PS I threw 3 of the needed lisp files into the EML "other" diectory in 
CVS for people to look over if they are interested.  You can get them 
from CVS or from the web here:
http://cvs.ecoinformatics.org/cvs/cvsweb.cgi/~checkout~/eml/other/physical-quantities.lisp?rev=1.1&content-type=text/plain
http://cvs.ecoinformatics.org/cvs/cvsweb.cgi/~checkout~/eml/other/standard-dimensions.lisp?rev=1.1&content-type=text/plain
http://cvs.ecoinformatics.org/cvs/cvsweb.cgi/~checkout~/eml/other/standard-units.lisp?rev=1.1&content-type=text/plain