[Voevent-core] Fwd: standard vocabulary

[Doug Tody]

Another pseudo-requirement to add to this would be that acronyms commonly
used in astronomy (GRB, AGN, QSO, etc.), be recognized in queries.
It would be fine to have more descriptive and precise names too, but
a good object-type resolver should accept such commonly used terms.
If the scheme already involves a one-to-many mapping capability on
input this should be no problem.

In this case I don't think we can defer all this to the user interface,
since a short term in an input query could respond to an arbitrary number
of lengthy descriptive UCD-like final classifiers, which is too cumbersome
to be propagating through a query interface.  This case is different than
that of a name resolver, where an arbitrary number of user-space object
names resolve to a single position on the sky.



On Fri, 26 May 2006, Doug Tody wrote:
> It appears that both VO-Event and DAL have similar requirements for a
> standard nomenclature for object classification, e.g.,
>
>    process.variation.burst;em.gamma
>    process.variation.burst;em.X-ray
>
> and so forth as has been suggested.  We had a similar discussion in
> DAL a while back but I will summarize the conclusions again here.
>
> In DAL (and probably VO-Event as well) this comes up in two different
> areas, object description via query response metadata, and in the
> query itself.
>
> For object description in the query response one wants as precise a
> description of the object classification as is known.  Long descriptive
> UCD-like descriptions as in the examples above are fine.
>
> For a query it is important that a more coarse-grained classification
> be possible, which allows a single string to potentially match many
> fine-grained type classifiers.  For example, "star" could match many
> object types.  It would be good if these coarse-grained classifier
> names could be brief for use in a query, as they will probably pass
> through to the service unchanged, originating from a human user or
> client application.  The relationships between the coarse-grained
> names and the detailed classifications could be one-to-many and
> possibly hierarchical as well.
>
> It would be good to avoid the use of special metacharacters in these
> names.  For example, in our earlier discussions, "*" was used as a
> shorthand for "star" in the naming convention proposed, but this could
> cause problems for software, for example if pattern-matching is used.
>
> We will need to standardize both detailed object classification and
> query input, so it is important to address both of these areas in
> whatever scheme we come up with.  One way to ensure that this happens
> might be to implement an "object resolver" service or web page to
> map short names to lists of classifiers once a list of detailed
> classifications becomes available.   - Doug
>