[Devforum] December 16th Meeting Notes

[Stephen Yearl]

All:

The PKI harvester is available here:
http://pkp.sfu.ca/pkp-harvester/

As I mentioned it's db structure (and record display) seems tailored for 
journal articles, so for it to be of utility for MSSA I will need to 
modify the code and the underlying database table structure. I'm still not 
sure that I want to do that, and there may be other better suited 
harvesters that we could employ instead. Pretty much out of the box, but 
with some skinning the PKI tool is running on an MSSA server here:
http://mssa.library.yale.edu/harvest/

drawing resources from MADID:
http://mssa.library.yale.edu/madid/oai/
and our local finding aids db:
http://mssa.library.yale.edu/findaids/oai.html

both of which data providers are employing the PHP code from:
http://physnet.uni-oldenburg.de/oai/

with modifications to accept a first-pass attempt at a genric "use policy" 
schema (feedback would be appreciated) here:
http://mssa.library.yale.edu/madid/oai/schemata/mssa.usepolicy.xsd
Content from this schema can be seen in the Identify verb issued to MADID:
http://mssa.library.yale.edu/madid/oai/oai2.php?verb=Identify

The finding aids OAI data provider works thusly:
1. local php code available from the finding aids admin pages traverses 
the dir of finding aids, stats the file system's modified date for the 
files and updates entries in a database if the date is more recent that 
that recorded in the db
2. for each file, the XML is parsed and sections are written to a MySQL 
database <titleproper>,<abstract>, <origination> and so on. The whole EAD 
instance document is also written as a blob to the db. This EAD MySQL db 
currently has no front end and exists solely to allow OAI data provision-- 
this just seemed easier to manage that a file system only based route 
(although Roy did mention that he found Open Source code available that 
works this way).

While the finding aids data provider says it offers oai_dc and oai_ead, 
requesting metadataFormat oai_ead is, I think, unweildy. For a single 
instance (e.g. 
http://mssa.library.yale.edu/findaids/oai/oai2.php?verb=GetRecord&identifier=oai:yul.mssa.ead:mssa.ms.1840.xml&metadataPrefix=oai_ead) 
things work OK, but a ?verb=ListRecords&metadataPrefix=oai_ead call just 
results in a PHP time-out as the size of the EAD instance documents are 
just too big. Another thing to note here is that there is no W3C schema 
for EAD, and http://www.loc.gov/ead/1.0/ 
http://www.loc.gov/standards/ead1.xsd in the sample GetRecord above is 
specious.

That being said, I am an the EAD schema working group, and I know that the 
next version of EAD will be issued in two schema (XSD and RNG) formats as 
well as the plain ole XML DTD. A part of this package will be an 'EAD 
fragment' schema that will allow for the embedding of EAD-based 
information in other schemata such as METS (or in this case the OAI 
response wrapper). FYI, this will essentially mean that <c> in EAD is made 
globally available as a root/document element in the schema.

Probably more than you wanted to know, right?

St.

On Mon, 19 Dec 2005, Karen Reardon wrote:

> Attending:
> Brian Kupiec, Art Bellanger, Jeff Barnett, Stephen Yearl, Jennifer Weintraub, 
> Julie Linden. George Ouellette, Matthew Beacom, Richard Boursy, Karen Reardon
>
> Stephen Yearl gave an overview of the use of OAI in MSSA. Current use with 
> MADID was explained. An overview of his intended use of PKI (Note: Stephen - 
> can you send along a URL?) as a basis for using OAI as a harvester to ingest 
> collections such as the Glee Club or the Yale Daily News was discussed.
>
> The discussion that followed led to the agreement that the Library should 
> standardize on a Common XML Response to a query that can be used across our 
> digital collections - Brian and Stephen will meet and come up with a proposed 
> document - we will discuss this at our February meeting.
>
> January Meeting Art Bellanger will give an over of Greenstone - and if 
> possible, its OAI implementation.
>
> Happy Holidays everyone!
>
> -kmr
>
> _______________________________________________
> Devforum mailing list
> Devforum at mailman.yale.edu
> http://mailman.yale.edu/mailman/listinfo/devforum
>