eric writes:
> Misha Wolf writes:
> >
> > I think this flags a logical problem in the DC definition. When we say that
> > "Each element is optional and repeatable", we aren't defining the relationship
> > between repeated elements. This can become problematical with some elements,
> > especially if we use a scheme (whether explicit or implicit). A related
> > problem is that HTML, or so I've read, doesn't promise to preserve the order
> > of META elements. This could have a serious effect on examples such as the
>
> On the subject of grouping, Michael Sperberg-McQueen wrote something very
> helpful after the Warwick conference that you may find quite useful...
>
> "On Information Factoring in Dublin Metadata Records"
>
> <URL:http://www.uic.edu/~cmsmcq/tech/metadata.factoring.html>
Michael's paper helps to clarify the problem with multiple instances
of a document, but does not suggest a general way of grouping
elements in HTML. He explains how, with 'and' and 'or' elements, you
can define a metadata record where it is clear which element applies
to which instance.
When embedding metadata in HTML, the instance problem does not arise:
the metadata is describing the current document. However, I think
that grouping is required, for example, to correctly associate the
names and affiliations two authors, in a machine-readable way.
It needs to be machine-readable so that you can do intelligent
searches on, say, name AND affiliation. Consider for example this
record, which has no grouping markup:
<meta name = "DC.creator.name" content = "Miller, Eric J.">
<meta name = "DC.creator.affiliation" content = "OCLC">
<meta name = "DC.creator.name" content = "Wicksteed, Charles">
<meta name = "DC.creator.affiliation" content = "Reuters">
This record needs to be marked with grouping information so that it
would not be returned from a search for:
DC.creator.name = "Miller, Eric J."
AND DC.creator.affiliation = "Reuters"
I am at a loss to suggest a simple syntax however :-)
Charles [log in to unmask]
|