The session started at a few minutes past 9:00 am with about 8 attendees. Dr. Newby led the session. The session proceeded as outlined in the planned agenda and online slides:

1. Welcome
2. Introduction of the chair and the agenda
3. Overview of IR, including why IR is well-suited for Grid computing
4. Status of GIR-WG documents:
   • Requirements document is published
   • Architecture document is underway, in draft form (expected to have significant progress by GGF13)
   • Specifications document is not yet started
5. Status of reference implementations:
   • MCNC released a technology preview
   • Nassar is proceeding with Amberfish integration to GIR (girtools), along with Newby's integration of IRTools
   • Newby & Nassar were TREC participants
   this year, and gained experience + insight into GIR applications for IR tasks
6. The session concluded with lively discussion among attendees about the real-world utility of GIR. While the indexing and collection management aspects of GIR are well-suited for the batch-oriented methods afforded by current Grid middleware implementations, the query function is not as well suited due to user expectations about the timeliness of responses to queries. For GIR, the value for queries might come from integration with desktop-based Grid-enabled systems which can operate standing or interactive queries. The benefit there would come from VO security considerations (i.e., for gaining access to particular datasets) and from standing queries (aka, information filtering).
7. The session ended shortly after 10:40

Notes by gbn September 21 2004