I wouldn't recommend trying to use the <structLink> facility
to do what you're trying to do, assuming that I understand it correctly.
It looks to me like you want two different views of the collection
of video files, one of which I would characterize as more-or-less
physical (here are the video segments composing the collection,
in order), and one more-or-less logical (here are the events and
scenes users will want to look at within the collection).
You are on the mark in using two structMaps, although I'd be
tempted to assign a TYPE attribute of "physical" to the first
(I'll concede the attribute value is a debatable point, and not that
important).
However, instead of trying to use <structLink> to associate the logical
structMap with the physical structMap, I'd recommend just creating
a more fully fleshed-out logical structMap. I can see two possible approaches,
and choosing between them would require knowing more about your particular
assets and their organization. If a scene always consists of one or more
segments, and the segment boundaries always match scene boundaries
(i.e., you're never going to have a scene which begins or ends in the middle
of a segment), then yes, I would do as you propose, a logical
hierarchy in the second structMap which would go:
Event
Scene
Segment(s)
<fptr>s
You wouldn't *have* to encode the segments to capture the information.
If you're first structMap indicates that segment1 = file 1 and segment 2 =
file 2,
then reasonably intelligent software (or humans) could look at a second
structMap that says scene 1 = file 1 + file 2 and figure out that scene 1
also equal segment 1 + segment 2. But adding the segment portion of
the hierarchy in the second structMap spells things out exactly and is
much quicker to process (if you need to know what segments go into a
scene) than pulling the information by inference.
If segment and scene boundaries don't align, I wouldn't bother putting
segment information in the second structMap. Just put in events and
scenes and relevant links to file information. Anyone wanting to know
how scenes and segments match up is just going to have to look at
the file pointer information in the two different structMaps and deduce
the relationships.
The <structLink> facility was really intended only as a mechanism to
record hyperlinks between assets that could not be appropriately mapped
into the hierarchical structMap (e.g., archiving web sites in METS). From
what you've described, it is possible to map your data using only structMaps,
so I'd stay away from using structLink as unnecessary extra work which
doesn't really add information.
At 02:16 PM 2/24/2004 -0500, you wrote:
>Hi,
>
>I am trying to use METS to represent a collection of video files. These
>video files are broken up into segments and I am representing this
>information as follows in a StructMap:
>
><METS:structMap TYPE="logical">
> <METS:div ORDER="1" TYPE="Collection">
> <METS:div ORDER="1" TYPE="Segment" ID="SEG1">
> File Pointer Information
> </METS:div>
> <METS:div ORDER="2" TYPE="Segment" ID="SEG2">
> </METS:div>
> <METS:div ORDER="3" TYPE="Segment" ID="SEG3">
> </METS:div>
> <METS:div ORDER="4" TYPE="Segment" ID="SEG4">
> </METS:div>
> </METS:div>
></METS:structMap>
>
>Further these segments are combined in no particular order to create
>"Events" (Just a term used for the purpose of my project), which are
>further divided into "Scenes". Every segment need not be part of an
>event. Hence I represent the Event/Scene hierarchy in a different
>StructMap as follows:
>
><METS:structMap TYPE="logical">
> <METS:div LABEL=" Ruth Stone's Collection " ORDER="1"
>TYPE="Collection" DMDID="DMRS" ID="Coll">
> <METS:div LABEL=" Gbarbea Funeral - Event/Themes/Voices
>" ORDER="1" TYPE="Event" ID="Event1">
> <METS:div LABEL=" Military Band Escort Body out
>of Monrovia " ORDER="1" TYPE="scene" ID="Scene1">
> File Pointer Information
> </METS:div>
> <METS:div LABEL=" Military Band Escort Body out
>of Monrovia " ORDER="2" TYPE="scene" ID="Scene2">
> File Pointer Information
> </METS:div>
> </METS:div>
> </METS:div>
></METS:structMap>
>
>So, now to represent what video segments make up an event, I use the
>StructLink area as follows:
>
><METS:structLink>
> <METS:smLink from="Event1" to="SEG1"/>
> <METS:smLink from="Event1" to="SEG2"/>
></METS:structLink>
>
>However, with this structure, I am not able to capture the sequence of
>segments that Event1 is composed of. Should this information be captured
>in the StructMap section and not in the StructLink section? To represent
>segments as divisions under events, doesn't seem right since it is not a
>strict hierarchy.
>
>Any feedback is appreciated.
>
>Thank you,
>Prajakta
Jerome McDonough
Digital Library Development Team Leader
Elmer Bobst Library, New York University
70 Washington Square South, 8th Floor
New York, NY 10012
[log in to unmask]
(212) 998-2425
|