[Dspace-general] DSpace data model

[MacKenzie Smith]

Hi Michel,

I'll try to answer this simply, and not too technically, since this is the 
general interest list. If you want the gory technical details we should 
take this discussion to the dspace-tech list instead.

-- Caveat: this answer is true for the 1.x releases only. The new 2.0 
architecture being designed now by the DSpace developer community has the 
concept of Archival Information Packages (AIPs from the OAIS reference 
framework for digital archiving) which would replace bundles completely.

-- The information architecture of Communities, Collections, Items, 
Bundles, and Bitstreams has always existed in the DSpace database tables. 
What we're discussing here is the degree to which the user interfaces makes 
it possible to put files into different bundles and treat them differently 
for display or other processing. What I will describe is how DSpace works 
*by default*, with the understanding that other solutions are possible with 
some local programming work.

-- In older versions of DSpace (prior to the current 1.2 release) the 
submission user interface automatically assigned all bitstreams to separate 
bundles and they were all treated the same in the end user interface. So 
the existence of bundles in the information architecture wasn't really 
being supported.

-- In the 1.2 release the concept of "named" bundles was introduced to 
handle the case of thumbnail versions of digital images, and full-text 
versions of documents (for indexing). In order to have separate named 
bundles for items, the submission UI was changed to put all the submitted 
bitstreams together into one bundle by default and to automatically create 
the separate thumbnail or full-text bundle as appropriate (i.e. the 
opposite behavior from before).

-- This means that there is currently no way *in the submission UI* to 
assign different uploaded files to different bundles. If you needed to do 
that, you would have to change the submission UI to support it. This means 
that the example I gave before (i.e. you want two bundles -- one for the 10 
TIFF images, and one for the PDF version) is *not* supported in the current 
UI, but it *is* supported in the underlying database tables -- you would 
just need to invent names for the new bundles, some way to assign uploaded 
files to the each bundle type, and some way for the public UI to know what 
to do with the different bundles (for example, display the PDF version, but 
suppress the TIFF version since browsers can't do anything with them).

I hope this is clear... it's pretty complicated stuff.

MacKenzie

At 03:24 PM 6/23/2004 +0200, Michel Auffret wrote:
>Hi MacKenzie,
>
>Thank you. It starts helping ;-)
>
>Could we go on with your example : how would you proceed to submit the "1 
>item / 3 bundles" you described ?
>
>Indeed, what I don't understand well is whether a "bundle" is conceptual 
>or real. Are the bundles manageable somewhere in Dspace or are they "only" 
>an internal packaging level ? In the administration tool I already met the 
>(sub-)communities, the collections, the files/bitstreams, but no bundle...
>
>Michel Auffret - CINES - France
>
>A 11:34 22/06/2004 -0400, MacKenzie Smith a écrit :
>>Hi Michel,
>>
>>"Bundles" are groups of bitstreams (i.e. digital files) that are closely 
>>related to form a complex digital object.
>>For example, if you scanned a 10 page document and created 10 TIFF 
>>images, then those 10 images should be in a bundle together since they're 
>>not meaningful as separate objects.
>>But if you also had a PDF file that was created from those 10 TIFF images 
>>then that could be in a separate bundle since it can stand alone. And if 
>>you also had a related dataset that
>>accompanied the document then it could also go in a separate bundle, 
>>since again it doesn't depend on either version of the document. But all 
>>three bundles belong to that one
>>"item", which is the logical work being archived.
>>
>>I hope that helps!
>>
>>MacKenzie
>>
>>At 03:15 PM 6/22/2004 +0200, Michel Auffret wrote:
>>>Hi everybody,
>>>
>>>We've installed DSpace here for evaluation purpose (looking for a long 
>>>term preservation software). I have a question about the data model. I 
>>>(think I) understand what a community, a collection, an item and a 
>>>bitstream are. But it's not clear to me what a bundle is. Could anyone 
>>>explain and/or give an example ? Are there some actions in DSpace that 
>>>are performed on bundles ?
>>>
>>>Thank you for your help,
>>>
>>>
>>>Michel Auffret - CINES - France
>>>
>>>_______________________________________________
>>>Dspace-general mailing list
>>>Dspace-general at mit.edu
>>>http://mailman.mit.edu/mailman/listinfo/dspace-general
>>
>>MacKenzie Smith
>>Associate Director for Technology
>>MIT Libraries
>>Building 14S-308
>>77 Massachusetts Avenue
>>Cambridge, MA  02139
>>(617)253-8184
>>kenzie at mit.edu

MacKenzie Smith
Associate Director for Technology
MIT Libraries
Building 14S-308
77 Massachusetts Avenue
Cambridge, MA  02139
(617)253-8184
kenzie at mit.edu