Jon,
C. M. Sperberg-McQueen did some work on information factoring
(grouping) and the Dublin Core that might interest you.
<URL:http://www.uic.edu/~cmsmcq/tech/metadata.factoring.html>
With regards to this, however, I'd like to suggest holding of
discussions on "grouping" for the time being and suggest getting back
to nailing down the "core".
eric j. miller <URL:http://purl.oclc.org/net/eric>
[log in to unmask] Office of Research, OCLC, Inc.
[log in to unmask] Dept. of Geography, The Ohio State University
Jon Knight writes:
> [Note this is basically a brain dump; I haven't prototyped any of this
> yet. Set your expectation sensors and flame throwers accordingly :-) ]
>
> A number have people have stated that we can rely on the ordering of META
> elements in HTML documents to be maintained by some software tools.
> Whilst I still regard these tools as broken, its a valid point and
> something we should consider bullet-proofing the embedded form of DC
> against. So here's my proposal: we introduce a "Grouping" qualifier. The
> value of this qualifier is a unique string for each logical group of
> elements. The physical ordering of the elements is then unimportant as
> the indexing software can match up elements to the groups that they are
> in. For example here's an extract of metadata from one of our HTML pages
> with the grouping qualifier added and the physical orderings rearranged:
>
> <META NAME="DC.author" CONTENT="(type=name)(grouping=jon) Jon Knight">
> <META NAME="DC.author" CONTENT="(type=name)(grouping=martin) Martin
> Hamilton">
> <META NAME="DC.author" CONTENT="(type=email)(grouping=martin) [log in to unmask]">
> <META NAME="DC.author" CONTENT="(type=email)(grouping=jon) [log in to unmask]">
> <META NAME="DC.author" CONTENT="(type=affiliation)(grouping=martin) Dept.
> Computer Studies, Loughborough University">
> <META NAME="DC.author" CONTENT="(type=affiliation)(grouping=jon) Dept.
> Computer Studies, Loughborough University">
>
> Note that Author is just one example of an element that could use this;
> I envisage Grouping as a qualifier that can apply to any elements and
> maybe even groups of different elements (so that you can specify the
> creation dates for instances of a resource in different formats and
> funky stuff like that).
>
> Now this does complicate things slightly for the metadata provider but:
>
> a) we need to do something like this if we expect our embedded metadata to
> pass through these tools unscathed and we're not going to make them have
> to refer to external metadata sources (which seems far more onerous to
> me),
>
> b) its completely optional and if you're only recording simple metadata
> you probably won't need to bother with groupings anyway (if you've just
> got a list of authors' names for example). This is what I expect most
> people will be generating, at least initially so this grouping qualifier
> is just to deal with the "power users", such as members of this mailing
> list!
>
> c) we do something similar to this with IAFA templates in ROADS (except
> its not quite as general because you can just group different elements
> together if you feel like it; I'm trying to learn from our experiences
> here!) and its seems to work OK for us.
>
> Anyway, I've got my asbestos underwear on so have at it people. :-)
>
> Tatty bye,
>
> Jim'll
>
> -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
> Jon "Jim'll" Knight, Researcher, Sysop and General Dogsbody, Dept. Computer
> Studies, Loughborough University of Technology, Leics., ENGLAND. LE11 3TU.
> * I've found I now dream in Perl. More worryingly, I enjoy those dreams. *
>
>
|