January 2011 Meeting - Pre-Conference Workshop Session 1 (CI and WS) Notes

January 2011 Meeting - Pre-Conference Workshop Session 1 (CI and WS and "others") on Content Interoperability Services in Bamboo Phase One 

9:00-10:00 am, International House on UC-Berkeley campus.

Bob Taylor (NU) – Introductions, about 25 participants F2F and 3 participants by audio/video conference.

Bruce Barton (WI) – overview of the goals of Workspaces & Content Interoperability "experiment(s)" during Bamboo Phase One.

Avoid developing API for obtaining content items of each collection, e.g.,
begin with a scholar that is collecting references to items in Zotero, then export item identifiers from Zotero and then bring items into workspaces. Workspaces has essentially a uniform metadata / content model of each object, incorporate remediation services as needed, ....

Bob – over the course of the production we may be able to move this work from experiment to demonstration to prototype (early production). At NU proof of concept with JCR has been done, so we believe this will work with CMIS for Bamboo.

Jonathan Smith (NU) – more about the details of how connectors work. Drew diagram. Concepts:

• Proxy for the content item in the local system, i.e. can act like local content, but remember where they came from.
• Adapters (CMIS) for each high-priority collection/repository
• CMIS – uses folders and documents, i.e., trees and nodes, ...
  

The object models of target collections vary considerably, e.g., XTF (CDL) relies on detailed XML files....

Connectors hide much of this variability – makes them look more like folders and documents.

In response to questions – the translated document maintains all of the original information, but fills out as required with a core object model, which supports sub-classing, etc. as operations are carried out.

Question – should these translation outputs be retained? TBD (in NU experiments during 2010 with JCR-based content interoperability, NU relied on caching). Maps of object must be cached and possibly retained, but the raw content (mostly) will stay with the source repository.

Bob – reactions from Workspaces and other CI. Enough and Sufficeint? Too much and Overkill? Is the CMIS path for content interoperability the correct one for this team's work in 2011?

Questions:

• Where is Discovery done?  (sometimes at the repository or collection.   sometimes a federated search of collections out of WS or BSP.  the content interoperability team deliberately does NOT want to impose a single practice model upon the larger Bamboo effort.
• What does the Workspaces have to retain? Derivatives, ....?
• Scale – what if a scholar needs to apply a Bamboo service over sub-corpora of 10,000 documents?

Steve Masover (UC-Berkeley) – this (CMIS framework) seems an appropriate way to start, certainly the model of bringing content over one at a time or a handful of objects makes sense for many use cases, but of course there will be other cases we’ll need to consider those use cases as well.

Patrick Schmitz  (UC-Berkeley) – how can we accommodate applying services close to the content? Does that require CMIS translators or the equivalent?

Tim Cole (UIUC)---there’s been some work on this with SEASR and similar projects. May be able to do some experimentation in Phase I, even if the heavy lifting on models for moving services to the content is mostly deferred to Phase II.

Bob -- Wrap Up points follow for what we are taking from today's pre-meeting.   Thank you for everyone's counsel, questions and help this morning!  

• The content interoperability team will continue to develop a CMIS-based interoperability plan for development work during the remainder of Bamboo Phase One.  
• This development work will be aligned, as much as possible, with the milestones of the Work Spaces teams for incremental releases in 2011 and early 2012 of the WS platform to actual scholars and students.
• This development work will be aligned, as much as possible, with the milestones and prioritized choices for collections/repositories from other focus areas of the CI effort and from the CS effort.   
• Allowing humanities scholars and students to get useful work done (in WS) with useful collections content is a MAJOR GOAL and ultimate success factor for the overall Bamboo Phase One effort.
• Follow up work must be done this week,  and also in follow up to the Berkeley meetings,  to develop: specific CMIS-based work milestones, beginning in February 2011; work assignments among NU, WI, UC-B teams, and maybe others; realistic estimates calculated for what effort level will be required to complete the work plan.