Comments to 1:1 rule
Another example is: an HTML-document with a little text and a photograph of
a painting of the coronation of the Danish King Frederik 3rd in middle of
the 17th Century
In the perfect 1:1-world there should be four sets of metadata:
1: Date for the HTML-page and Creator of the HTML-page etc.
2: Date for the photograph and Creator of the photograph etc.
3: Date for the painting and Creator of the painting etc.
4: date for the event (the coronation), a Contributor (The King?) etc.
But in the real world there is only one set of metadata: the set on the
HTML-page. These set of metadata should include a Description of the event -
it is the object of the HTML-page.
In a database of a museum with the painting on the wall you may be will have
a data structure with four sets of metadata - but not on the Internet. You
can't use the 1:1 principle as a general principle for embedded metadata.
Regards,
Leif
********************************************************
Leif Andresen * Email: [log in to unmask]
Library Advisory Officer
Danish National Library Authority
Nyhavn 31 E, DK-1051 Copenhagen K
Phone direct: +45 3373 3354
Phone: +45 3373 3373 * Telefax: +45 3373 3372
********************************************************
> -----Oprindelig meddelelse-----
> Fra: [log in to unmask] [mailto:[log in to unmask]]
> Sendt: 26. juli 1999 20:35
> Til: [log in to unmask]
> Emne: Re: perfect and less perfect worlds
>
>
> Robin -
>
> I've never seen a succinct definition of the 1:1 rule. Those
> of us who
> came in after the original discussions have had to piece it
> together from
> context ... (_please_ someone whose got a better
> understanding jump in at
> any time) ... but ...
>
> As I understand it, the 1:1 rule deals with only including
> metadata that
> applies to the instantiation of the resource "in hand". It is usually
> thought of in terms of related resources. Hence, if a resource has a
> relationship with a second resource. it may reference the
> second resource,
> but it should not include metadata from that resource (unless
> that metadata
> is also directly applicable). If people (or
> applications/search engines)
> want more information (metadata) about the related resource, they must
> follow the link to that resource.
>
> For example, suppose we have a resource called "Article A" by
> Suzy Smith
> and a second resource called "Article B" by John Doe. If Article B is
> based on Article A, as I understand it, the 1:1 rule tells us that we
> should indicate the relationship in Article B's metadata:
>
> DC.Title = "Article B"
> DC.Relation.IsBasedOn = "Article A"
>
> but that we should stop short of including any other Article
> A metadata.
> For example, if you want the author of Article A, you would
> have to go to
> Article A's metadata to get it. You would _NOT_ add an
> Article B metadata
> field like DC.Relation.IsBasedOn.Creator = "Suzy Smith".
> Simply put, even
> though there is a relationship between the resources, Article
> A's author
> does not belong in, nor should it be repeated in Article B's metadata.
>
> So what's the parallel to, for example Creator.EmailAddress?
>
> From a conceptual perspective, the agent qualifiers can be
> thought of as
> special cases of the Relation qualifier in so far as they
> name a related
> resource where the relationship is that the named resource created,
> published or contributed to the current resource. For
> example, the DC 1.0
> notion of Creator could conceptually be expressed as
> Relation.WasCreatedBy,
> (if there were such a subelement). Hence
>
> DC.Title = "Article B"
> DC.Creator = "John Doe"
>
> could be represented as
>
> DC.Title = "Article B"
> DC.Relation.WasCreatedBy = "John Doe"
>
> Hence, adding DC.Creator.EmailAddress would be tantamount to adding
> DC.Relation.WasCreatedBy.EmailAddress (like
> DC.Relation.IsBasedOn.Creator).
>
> John Doe's e-mail address is part of the metadata that
> describes the second
> resource Johh Doe, not part of the metadata that describes Article B.
> John's e-mail address has no more business in Article B's
> metadata than the
> Article A's author does. (IMHO)
>
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|