In attendance:

(in no particular order)
Bernie
Steve
Bill
Pib
Tim Cole
Martin
Duane
Amit

Topics of Discussion

Regularizing teiHeader using XSL stylesheets for keywords, sourcedesc and various other metadata holders.

We agreed that regularizing teiHeader would be good thing to do, Bernie had good points about doing this process
inside the datastore in triple store. Martin, Bill, Steve and others thought that it is always beneficial to have a representation that humanist scholar can look in an editor and XML as a processing medium is very flexible compared
to ingested texts in the database.

We decided that Steve, Martin and Bernie would work together and give recommendations to the data cell. The initial
effort could be to look more concretely to the Bibliographic elements that have been identified on the wiki.
Web URL?
Time line?

Metadata scheme that we can adopt for collection metadata search.

Tim suggested we look at the Dublin Core for collections. This scheme would support discovery and browse of the collections.
Tim also mentioned that dublin core for collections also has elements that describe the relationships between the collectons.
Tim writes

Relative to discussion on today's conference call regarding
collection-level descriptive metadata schemas:

Here's the URL for the Dublin Core Collection Description Application
Profile updated and reviewed (and declared "conformant") by the dcmi
usage board earlier this year:

http://dublincore.org/groups/collections/collection-application-profile/
index.shtml

with a cliff notes version at:

http://dublincore.org/groups/collections/collection-ap-summary/

see also:

http://dublincore.org/collectionwiki/

and 

http://dublincore.org/groups/collections/index.shtml

Amit volunteers to work with others on this topic
TimeLine? This needs to be integrated with the workflow collection ingestion process, so a prototype proposal would be
in order with in a few weeks

Ontology -classification, attributes, relationship between collections and how they should be represented.

Pib pointed out that we should be careful since this is subjective and usually user specific. There was a general consensus
that we need some level of classification and Martin pointed out that we could perhaps use LC numbers that are assigned to each text. Tim mentioned that LC numbers don't go very far for prose and there was a consensus that it would work out and we will explore this issue.

_ Kelly, Tim and Brian would lead this effort in consultation with collection curators._

Authoritative copy of the teisimple.dtd

-Amit and Bill will put the copy of the dtd in SVN.

Timeline: By the end of the week of Aug 02

Dealing with not-quite-TEI collections in which data demanded by teisimple is not expressed at all

Next Week:

Formalizing workparts and chunks across the collections for data mining.

Next Week:

How these activities should be integrated in the workflow.

Next Week: