Some thoughts to throw into the mix on this:
 
Metadata vs Sequence
 
I'd suggest that for instance page numbers are metadata (about individual
pages, what's printed on the page), from which sequence could be implied,
rather than a direct manifestation of sequence.  Some (perhaps esoteric)
examples where using page numbers directly might be
- differing page number schemes within a book (i, ii, iii in the
introduction; 1, 2, 3 in the main body etc)
- missing pages (usually unlikely, but for instance if you were digitising
the only copy of a resource available, and it was incomplete, you couldn't
rely on page numbers for "page turning" in an application without some
additional logic)
- errors in either the page number order or the sequence of pages in the
original resource (again unlikely, but...)
So there's an argument for expressing both things like page numbers as
metadata, and sequence in terms of before/after/next/previous relationships
(perhaps having rules to derive one from the other).
 
Sequence of resources vs sequence of relationships
 
Sequence can be thought of as a property of the relationship between the
resource and collection (page and book) rather than a property of the
resource itself.  That is to say that one page follows another only by
virtue of the pages being in the book - having a relationship to the book.
There could be cases where a resource was in different collections and you
wanted to express resource sequence differently in different collections -
so the sequence is strictly a property of the relationship itself rather
than the resource.
 
Where the relationships are "stored"
 
There's cases for both storing relationships in a parent object (the book
itself), or in the components (pages), I think this probably comes down to
manageability depending on the scenario.
 
rdf:Seq
 
>From [1], rdf:Seq .. "is no different from an rdf:Bag or an rdf:Alt" and
"the rdf:Seq class is used conventionally to indicate to a human reader that
the numerical ordering of the container membership properties of the
container is intended to be significant" - so I'd suggest that semantics of
rdf:Seq are possibly not that useful (leaving aside the implementation
problems!).
 
Regards
Steve
 

-----Original Message-----
From: Gregory Jansen [mailto:[email protected]] 
Sent: 19 May 2010 14:32
To: [email protected]
Subject: Re: [Fedora-commons-users] Object Order Using RDF


Hi Andrew,
Like Richard we preferred to encode our parent-child relationships in the
parent object.  This makes it much simpler to reorder or insert elements
later.  

I looked into embedding an RDF:Seq in RELS-EXT, but this is not possible.
RELS-EXT RDF is restricted to be one level deep, with no blank nodes, as I
understand it.  A Seq required creating a new node in the RDF for the Seq
resource.  You might be able to achieve this today with the RELS-INT
datastream, but I'm not up to date on any restrictions in that RDF.

The other trade off is simplicity.  RDF collections are unwieldy in my
opinion.  I do put a parent-child relationship in RELS-EXT, but I do not
record the sequence there.  The sequence is in a separate METS structMap
datastream.

Greg

____
Gregory Jansen
University Libraries
University of North Carolina at Chapel Hill

On 05/19/2010 08:13 AM, HIGGINS R.I. wrote: 

Out of interest, why not use METS to encode this relationship at a book
level?



As always with FEDORA there seem to be several ways to do things, so it is
always useful to know why one particular one has been chosen. I find that a
METS file for the book level digital object stores all the metadata in one
place which simplifies reuse by applications. Although you can never
eliminate getting data entry just plain wrong in any format, I could see a
situation with the RDF solution where one mistake would mean you jumping
into a different book altogether (as opposed to the file not found errors I
get when I get the METS file wrong ...)

- - - - -
# Richard Higgins
# Durham University Library
# Archives & Special Collections
# Palace Green
# Durham
# DH1 3RN
# E-Mail: [email protected] 



From: Andrew Curley [mailto:[email protected]] 
Sent: 18 May 2010 22:17
To: [email protected]
Subject: [Fedora-commons-users] Object Order Using RDF



Does anyone have experience using RDF to describe sequential relationships
between objects?  For example, given two objects representing two page
images, I wish to describe that one object precedes the other and/or that
one object follows the other.  There is nothing in the Fedora ontology that
facilitates this relationship so I was wondering if anyone in the community
has codified this relationship through RDF or knows of an RDF standard which
facilitates this description.



Thanks,

Andrew Curley



----------------------------------------------------------------------------
--



  


_______________________________________________

Fedora-commons-users mailing list

[email protected]

https://lists.sourceforge.net/lists/listinfo/fedora-commons-users

  


------------------------------------------------------------------------------

_______________________________________________
Fedora-commons-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/fedora-commons-users

Reply via email to