Supercell call October 1, 2007

Present: John, Matt, Bill, Stan, Catherine, Martin

Topics:

#1: Collaborative behavior across groups/sites. Esp. w/ respect to use of SVN, related collaborative tools. Bill will talk to Amit, then to PIB, about this.

#2: Role of the proxy group, which might relate to definition of proxy calls vs. asset actions (or the larger question of how does Fedora impact the proxy) – Thursday fedora discussion... Bill: Fedora disseminator (URL Syntax that includes fedora base URL; get function; id of object; behavior definitonID; method invoked) shouldn't be the public face of MONK; we should stick with proxy or move to asset actions (an attempt to abstract some of the modeling done with fedora by some institutions (UVa, Tufts, Northwestern) and express access points in repository-neutral ways; an xml structure that has action-elements, actions have labels and URIs). Get-label, get-preview, get-chunk, etc.; extending the list of action-elements is a community process; we could propose a monk set for quantitative access to texts. Consuming applications process asset actions, for example OAI harvesters and collection-builder/image annotation; Collectus tools at Virginia. For text, the question has always been 'what's the consuming application' beyond skinning, etc.'. Proxy might be consuming application; should asset actions be the langauge the proxy speaks? Good time to mention that Amit, Martin, Catherine have revised lexicon, let's focus on proxy/middleware/etc. and bring it in line with the architectural diagram and our understanding of what is in each part. Proxy API needs discussion too: does it migrate the expression of this to asset actions; parts are solid, parts are speculative.

#3. Metadata decisions: "in" the data cell as of the most recent notes, but also spans all other groups in some way. Metadata requirements should be produced from other cells for the data cell, based on use cases, in an iterative case-by-case process. Other cells should perhaps look at several cases before forwarding the requests, to make sure that there's not unmotivated difference across cases. Make it clear where data cell looks for it. Make it a very simple list.

#4. Is choice of a schema langauge still a live issue? TEI-Simple? Nope: schema language as you like; TEI-Simple pretty much as it is.

#5. Exposing collections: how? When? As soon as Brian has adapted his xsl to tei-simple, Wright can go through the hopper, morphadorning (with training data for American NCF), NCF and Wright would go into Fedora. Thanksgiving, we'll have collections we can browse.