*Attendees*
Scott Cantor, OSU
Tom Barton, U. Chicago
Grace Agnew, Rutgers
Steve Carmody, Brown
Bob Morgan, U. Washington
Art Vandenberg, GSU
Fred Beshears, UC-Berkeley/IMS
Bob Morgan, U. Washington
David Bantz, U. Alaska
Jeanette Fielden, Internet2
Renee Frost, Internet2
Steve Olshansky, Internet2

*Discussion*
The Birds of a Feather (BoF) at the Internet2 Fall Member meeting is scheduled for Monday Oct. 13th at 10:30 a.m. The BoF will discuss definitions, scenarios, use cases, and present the strawman courseID structure for review. Grace suggested that it would be very useful to hand out blank diagram structure sheets and have people fill them out with course structures at their institutions. It would be beneficial for vendors to participate in the BoF. Subject matter experts can attend the BoF at no charge but if they want to attend the full Internet2 Member meeting they will have to be a paid registrant. BoF materials are time urgent since they will have to be finalized on the September 29th call.

There is an area on the web site for draft documents being developed by the CourseID group. Documents not being developed by the CourseID group can be listed under the links section to ensure that the most current version is being referenced on the document owner's site.

IMS creates specifications not formal standards. Organizations like OKI and ADL will take the specification and make them specific enough for implementers to begin testing them. A specification might then be submitted to standards group such as IEEE or ISO for ratification. The courseID use cases with Shibboleth would be in the implementation area. Internet2 does research and testing and then, if needed, may submit results to a standards body for ratification. CommObject was developed and tested by Internet2, then submitted to the ITU and adopted as H.350. Other efforts such as eduPerson are higher education specific and don't map readily to a standards organization and are managed by MACE. Scott's concern is: What is the right way for Shibboleth to be incorporated so that it doesn't become just yet another mechanism for people to deal with but a solution that makes things easier to accomplish? CMS by their nature are tracking information and associating it with people. The use case proposes to do in real time what the IMS enterprise spec purports to do in batch: provision users into the system so they have access when they try to log in. The purpose of the discussion is to determine if that should be pursued.

The IMS Resources List Interoperability group is working on the problem of how you handle reading lists since inserting a URL in a CMS will often break, usually due to access control issues. Elsevier and Blackboard have worked out a solution between their two systems but it's not standardized. (http://www.imsglobal.org/rliCharter.pdf). So there is an interest in the tie-in to the library world and how to deal with digital object identifiers or URN's for digital objects and how to authenticate/authorize people to access those digital objects for multiple campuses.

There is project at Harvard working on a more general solution for independent resource URL's and how to keep them from breaking across systems. It's a technically difficult problem and library journal vendors would have to be persuaded to upgrade their systems to enable access to online journal articles in a way that doesn't break links. Till then you need to cope with the ad hoc solutions.

Fred has mentioned the CourseID work to the IMS Enterprise group and they are interested in how multiple campus situations would work. Vendors such as WebCT are using the Enterprise spec to run websites. The course management systems exist because registrars haven't wanted to deal with all the variations in the courses and enrollments that TA's and others need to work with. How would the student information system communicate to the course management system who is in the class and authorization for access to materials? How could Shib be used to facilitate initial rostering, with the CMS relied on to handle finer grained things like exam results? Would a phase one goal be not worrying about how students at Berkeley are added to UCLA but worry only about them being manually added to system at UCLA and how that is done?

Fred suggested that as a first step the UC system might be a good test for cross campus access. IMS has the concept of source id. For Berkeley, there is the course control number, a number string that uniquely identifies each lecture, lab, discussion section within a given semester. So a combination of an identifier for the Berkeley registrar and the course control number would uniquely id each section within the 2003 spring time period. Start with building on the source id for an offering and then determine if it's desirable to build a class schedule directory. The directory could offer ids for multiple campuses so if a student needs access to a course site at another campus, Shib could be used to make that more effortless. Procedural issues regarding registrars and course credit could be regarded as outside the immediate scope.

Grace felt the id should recognize the entity relationship. The course id attribute should not preclude other attributes and to ensure inter-operability there would need to be awareness of what the other attributes would be.

Bob asked if any information has been collected about actual practices at universities. It was agreed that a survey might produce insight into how universities identify courses and use those identifiers. Educause might also have information of relevance. The School of Information Science at Berkeley worked with the SIS group to identify student roles, schemas etc. so examining their report would be beneficial.

There was general agreement that while other functionality is not meant to be excluded and might be developed later only what is specified in the use case will be supported initially. What it won't support can be more concretely defined at a later date when there is a better understanding of what we can do. Fred recommended a book called Writing Effect Use Cases by Alistair Cockburn published in 2000 ISBN 0201702258.