[OAI-implementers] dc:title vs. dc:description for images
Sarah L. Shreeves
sshreeve at uiuc.edu
Mon Aug 30 11:26:16 EDT 2004
I've replied to Jenn on another list about this question (so I apologize
for anyone who's getting a duplicate response!), but I thought it would be
useful for other service providers to respond as well...
As a service provider we've actually made a point of making some rules
about displays and indexing that take into account issues like no titles.
For instance, in our Mellon funded service provider (for cultural heritage
materials), we always display the dc:title, dc:creator, and dc:subject or
dc:description in our short display. If the dc:title is missing, the
dc:subject is displayed. If dc:subject is missing, dc:description is
displayed. We also included the title, subject, and description fields in
one index. This decision was made primarily because of the amount of
museum (and library, for that matter) metadata no titles. In our study of
the use of DC by community (from the Mellon service), we found that 77% of
records coming from museums and 66% of records from academic libraries had
a title element. On average, records from the museum community had one
title entry per record, and those from the library community had about two.
So I think that either Option 1 or Option 2 would work fine. Service
providers who are working with non-bibliographic material at least should
be accustomed to metadata that doesn't fit the normal mold. This doesn't
mean poor quality metadata, of course, just that 'good' metadata doesn't
have to have a title field!
Also, as you note, Option 4 is absolutely the worst option from a user's
point of view!
At 12:50 PM 8/28/2004, Riley, Jenn wrote:
>I'm mapping metadata for an archival collection of digitized slides (the
>collection is at <http://www.dlib.indiana.edu/collections/cushman/>)
>from our native format to simple DC for exposure via our OAI data
>provider, and I'm having trouble making a final decision about how to
>deal with a dc:title vs. dc:description issue.
>These slides are from a skilled amateur photographer, so they do not
>have a "title," per se. They do have a very large amount of description
>however, all provided by the original photographer - a description
>written on the slide mount, plus similar but not identical descriptions
>in between 1 and 3 small notebooks accomanying the slides. So we have
>between 2 and 4 descriptive strings for each image, none of which can be
>considered to be more authoritative or complete than another. I'm trying
>to decide whether to put these strings in dc:title or dc:description
>Option 1: Put all of them in dc:title. This means each DC record will
>have up to 4 dc:title elements. I'm leaning strongly toward this
>solution at the moment, mainly because I don't find any of the other
>Option 2: Put all of them in dc:description. This is from a pure
>metadata standpoint the best solution - these strings are all really
>descriptions, rather than formal titles. However, this would result in
>NO dc:title element for these records. Since service providers rely so
>heavily on dc:title for intelligible display of records, I'm extremely
>hesitant to use this solution.
>Option 3a: Pick one of the strings to put in dc:title and put the rest
>in dc:description. If we were to do this, I suppose the description from
>the slide mount would go in dc:title, since there's only ever 1 of
>these, but I really don't like this solution. There's no reason to
>prefer this description over the ones from the notebooks - in fact, the
>descriptions in the notebooks were written when the picture was taken
>(the slide mounts obviously were written after the film was developed!)
>and the notebook descriptions weren't subject to space considerations as
>the slide mount descriptions were. But there's no reason to pick any of
>the 3 possible notebook descriptions over any of the others so there's
>no good way to pick one of them for dc:title either.
>Option 3b: Pick one of the strings to put in dc:title and put all of
>them in dc:description. Same as above, but duplicate the value in
>dc:title in dc:description. I don't like this solution for the same
>reasons as 3a, above.
>Option 4: Create a generic title for each record (something like
>"Charles W. Cushman slide") and put all the descriptive strings in
>dc:description. Although I'm a librarian and we construct "titles" for
>unpublished materials all the time, I *really* don't like this solution.
>There are 15,000 slides in the collection, so if a user's search came up
>with more than one record from this collection, a display listing titles
>would be extremely UN-useful. Also, we have a strong commitment in this
>collection to preserve the photographer's original words, so I'd like to
>find a way to make these what users see in search results.
>I do plan to expose at least one other more robust metadata format for
>these records, so hopefully some service providers will be able to see
>and maintain the complex nature of this issue, but I still want to make
>sure that our simple DC is intelligible for service providers who will
>only use that metadata format.
>So does anyone out there have advice on this issue? Is there a pervasive
>practice out there by service providers to use a dc:description for
>display when there's no dc:title? What have other data providers done
>when dealing with unpublished material with multiple possible "titles"?
>Digital Library Program
>Indiana University - Bloomington
>Main Library E170
>OAI-implementers mailing list
>List information, archives, preferences and to unsubscribe:
More information about the OAI-implementers