Howard - I've taken a very quick look over your DIG35 metadata standard.
My immediate reaction is that the primary area that we may need to look at more
closely concerns the Content Description Metadata given in Annex D of your spec.
Most of the metadata in the other sections is either beyond the scope
of DCMES, or can be mapped quite straightforwardly to DCMES.
The latter point is probably important: DC does not intend to
"rule-the-world" of metadata, but rather to provide a lingua-franca
for exchange, with a likelihood that it can act as a lowest-common-denominator
for public indexing purposes. Thus, one-way mappings *from* more specialised
schemas *to* Dublin Core are the most likely direction of movement, to allow
union-catalogues to be presented to users.
In DCMES there are probably 5 elements that primarily concern the
"content" of the object being described:
- Title in most cases is a potted description (though not always!);
- Type, for digital images will normally be "image" at the highest level
(though there have been some proposals for greater granularity, so we may
find a vocabulary including terms such as "portrait", "group", "landscape",
"landscape-rural", "landscape-urban", "still-life", etc etc developed);
so it looks like the remaining three from DCMES which will have to do the
bulk of the work are
- Description (which is effectively a default location for content-description),
- Subject (i.e. keywords) and
- Coverage.
Considering the DIG35 content-description:
- Image Overview/caption is DC Title.
- Comment is text that could only be mapped to DC Description.
- Keywords can go into DC Subject.
- Audio specifies a related resource, which would normally be the target
of a DC Relation element.
- Dictionary is a 2nd order descriptor which is called a "value-qualifier"
or "value-encoding-scheme" in qualified DC.
After that, in the DIG35 standard it appears that you have made some judgements to
promote particular typed-descriptors to the top, which is advantageous in the likely
application domain. It is partly because of this that it is not clear that there
is a straightforward one-to-one mapping from these to DCMES.
- Organisation and Keyword-set both concern the general layout - this kind of structuring is beyond what is normally found in DCMES
descriptions.
- All of Person, Thing, Keywords might be used to isolate specific items found in an image, and will usually correspond to DC Subject, with some
loss of semantics.
- Location and Event both indicate context - Location fits nicely in DC Coverage (the one typed-descriptor in DCMES ...), but Event is not
handled discretely in DCMES: it is probably most cleanly managed in DC by attaching the Event to the description through a DC Relation, but
event related information also commonly resides in Date.
As implied above, it is pretty much always OK to stuff otherwise-homeless information into DC Description.
And if you want to go to the trouble of adding qualifiers to DCMES then you can do this precisely (though the qualifiers may not be useful to
anyone else ...)
--
Best Simon
|