Hi Roy:
Thanks for taking the time to explain all that. A few comments below.
Roy Lowry wrote:
> Hello John,
>
> Mapping CF Standard names to GCMD Parameter Valids to the public
> domain 'variable' level shouldn't be a problem as the GCMD vocabulary
> is clearly a discovery vocabulary. The issues Bryan is eluding to
> result from trying to map a usage vocabulary (one containing terms
> that fully describe a data value) because vital information for the
> mapping (e.g. whether a value is a mean or a standard deviation) are
> held in CF fields other than the Standard Name.
>
> Are you aware of the MMI vocabulary mapping workshop that was held in
> Boulder last August? CF was one of the vocabs used, but the work
> really only scratched the surface looking at maybe half a dozen
> terms. There should still be an OWL version of CF on the MMI site
> (as well as GCMD), but this will a snapshot that is now well out of
> date.
> There is however an open source tool to convert the XML to OWL
> (voc2OWL) and a very neat mapping tool (VINE). The resulting map is
> an OWL file that can be used with web services also provided by MMI
> to build a thesaurus server. I would strongly recommend their
> approach.
I have grabbed the files at
http://marinemetadata.org/2005/02/cf
http://marinemetadata.org/2005/02/gcmd
they are OWL XML files. Are these the ones you are recommending?
>
> The one issue with the GCMD parameter valids that was never properly
> sorted at Boulder is that the mapping was done to GCMD variable terms
> without their overlying term hierarchy, which took no account of the
> fact that EARTH SCIENCE > Biosphere > Animal Taxonomy > Fish and
> EARTH SCIENCE > Oceans > Marine Biology > Fish are different. I
> know Luis Bermudez of MMI was looking at this issue, but don't know
> how far he got.
>
> Building an ontology mapping between CF and GCMD was on my agenda for
> later this year to provide the capability to NDG for entering a CF
> Standard Name into a discovery portal and finding DIFs marked up
> using GCMD parameter valids. But I'm more than happy for someone
> else to make a start and would of course be happy to help once the
> time slot I've allocated for later this year comes around.
Yes, I need
CF standard name -> GCMD parameter valid
and also
Grib1/Grib2 parameter -> GCMD parameter valid
I have a simple start on the latter, courtesy of Gene Major at GCMD.
Im unclear on whats all needed to implement the service you describe?
>
> Cheers, Roy.
Thanks!
Received on Thu Apr 06 2006 - 16:28:59 BST
