[kepler-dev] EML Unit information in Kepler

[Matt Jones]

Hi Dan,

Yes, I think it is a feature we should expose.  We need a general way to 
expose all of the metadata via the actor.  It would be easy enough to 
make it available on a 'metadata' port, but this is a bit of a hack -- 
we really need to parse the metadata and make it available in a more 
structured way consistent with how Ptolemy would expect.  For example, 
Ptolemy's ports can have unit assignments, so we could be modifying the 
ports themselves to reflect the units.  We need to discuss how to handle 
this, and as Shawn indicates, we probably need to revamp how data is 
managed in Kepler altogether.

Matt

Dan Higgins wrote:
> Hi Matt,
> 
>    During my recent discussions with Akiko Ogawa, a limitation in the 
> current EML2 dataSource actor came up. The basic problem is that with 
> the current implementation, there is no way to programatically  get the 
> units of attributes! The ports are automatically created and type 
> information (e.g. ints, doubles, etc) assigned, but Akiko wanted to have 
> the units of each attribute available so that unit conversion could 
> automatically take place. [The hope was to have actual unit info and 
> desired unit info available as inputs to an actor that could convert the 
> actual units to the desired unit for output to some model. This would 
> involve reading the actual unit from EML and having the actor figure out 
> how to convert it to the desired unit and then apply that conversion to 
> all of the data referred to be the EML attribute.]
> 
>    Clearly this unit question is similar to that for any type of 
> metadata. How do we do automatic conversion? Units are iust a obvious 
> case. Unfortunately, our EML2 DataSource actor does not provide any of 
> the unit information.
> 
>    I am not at all sure how we should/could provide this unit 
> information. Will it come with the semantic annotations tied to 
> measurement ontologies, or is it a special case that we could handle 
> some other way (especially for EML2 actors where the information is 
> already in the EML metadata). I think it is a feature that Akiko would 
> like to see in Kepler as soon as possible, so I a am suggesting that we 
> give it some consideration in the near future.
> 
> Dan
> 

-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Matthew B. Jones
jones at nceas.ucsb.edu                         Ph: 805-617-4179
National Center for Ecological Analysis and Synthesis (NCEAS)
UC Santa Barbara     http://www.nceas.ucsb.edu/ecoinformatics
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~