handling metadata with multiple values
Wil O'Mullane
womullan at skysrv.pha.jhu.edu
Tue Aug 12 11:46:08 PDT 2003
I did not like the fact that in your schem everything was
<string>
e.g.
you would have
<ucdList>
<string> ucd1</string>
<string> ucd2</string>
</ucdList>
I do not like this.
But nothing wrong with (and you mentioned in your own mail
<ucdList>
<ucd> ucd1</ucd>
<ucd> ucd2</ucd>
</ucdList>
Which is the best of both worlds - or at least thats what I read from Anita
On Tue, Aug 12, 2003 at 01:33:34PM -0500, Ray Plante wrote:
> Hi Gretchen,
>
> Hmm, I thought you guys didn't like the list elements (or at least Wil
> didn't).
>
> > I also endorse the ucdList element and hope to see this added into the
> > VOResource schema. This element would simplify mining metadata content
> > to find potential data sets which will satisfy data value driven query.
> > In other words a high level filtering capability.
>
> Could you expand on this a little more about how it would help? That is,
> how do you envision extracting values to test against a query? (e.g. will
> you extract into a database first? using an ots tool or something custom
> written on top of, say, DOM & JDBC?)
>
> There is no difference in the information stored in the two approaches.
> So the differences are in how you access the data programmaticly and how
> it looks. (Any others?) My discussion addressed the programmatic issues.
> I don't want to discount the ascethetics, but I think it is important to
> understand which kind of issue we're talking about.
>
> cheers,
> Ray
>
>
>
>
More information about the registry
mailing list