Re: [stdsreg] Metadata Standard v2.5

[Martin Bryan <mtbryan@sgml.u-net.com>]

Bob

I think the paragraph you questioned in yesterday's meeting was the
following one:

> Should we require that Referenced Standards entries be entered as sets of
> Identifiers? The alternatives seem to be Designation and Title, one of
which
> is optional and the other long but mandatory. Should we make Identifier
> mandatory? Or should we assign records some form of mandatory "key" which
> can be used in Referenced Standards?

The question I raised is "What from should a reference take? Should it be:

a) The mandatory full name of the standard
b) The standard's optional designator
c) Some form of key to the metadata entry for the standard?"

The underlying question is one of consistency. Given that different people
may be referencing a standard from different organizations, how can we
ensure that names or designators are entered correctly or up-to-date? For
example, the name of the W3C RDFS specification has changed recently. Any
reference to the old name will no longer satisfy a search of the metadata if
it has been updated to contain the new name. Also, what happens when ISO
13250:1998 is changed to ISO 13250:2002 when a second edition is published?
Unless both versions of the metadata record are retained there will be a
mismatch. If both records are retained, how do we indicate that one version
has been supersceded by another?

Ideally any reference should be in the form of something that is
"navigable": i.e. once you get to the reference clicking on it should
generate some form of pointer to the record for that standard within the
relevant metadata repository. Whch brings me to the other point I made last
night. Given that we have a set of metadata records conforming to our
proposal stored at different sites by different SDOs, how can users search
for a record? We need some mechanism for identifying currently available
sets of records, and a mechanism for distributing queries to all existing
record sets. This mechanism should be able to produce a single response from
multiple replies for those cases where searches are based on keywords, etc.

Martin Bryan