IW Meeting 2007 July 19

From Inference Web

Jump to: navigation, search



  • UTEP: Paulo 915-747-6373
  • RPI: Li 518-276-2526
  • Stanford: Deborah, Cynthia, Alyssa


  • Announcements and schedules (Li, Deb)
  • GILA report (Li)
  • TAMI report (Li)
  • PML2 code consolidation report (Cynthia)
  • Summer student's progress (Cynthia)




  • Deborah: AAAI travel (July 20 -July 31), rpi trip (aug 4-aug 7)
  • Cynthia: vacation 14/08 - 28/08
  • Li back on July 24
  • Li leave after Aug 17
  • glass gone july 21 - july 31


  • July 25 1-3pm PST (call Deb at Canada, paulo at utep)
  • Aug 2 2:30pm iw meeting on GILA/TAMI (alyssa will come late)
  • Aug 8 11am pst (call paulo at utep)
  • Aug 16 1:30 pm pst (call paulo utep)


  • meeting procedure
    • meeting agenda before meeting, put it on wiki
    • dump IRC log to discussion page of Wiki
    • produce action items after meeting, put it on wiki


pml2 code consolidation(cynthia)

  • clearify user requirements, what pieces we are going to feed our users
  • make sure we have consistent source code development strategy

gila reports

  • No work before September
    • Oct 07: new SDK
    • Dec 07: port to new architecture; collect new requirements
    • Feb, March, May 08: integrate phase 2 features
    • July 08: debug, stabilize, prepare system
    • Aug 08: final drop
  • Ken and Martin reviewed Darpa report
    • guidelines have been re-emphasized
      • Learning from one rich example (not 1000 examples)
      • Learn to choose best deconfliction step (not planning best solution)
    • Making application more realistic
      • Learn acm dependency from demo (TowD)
      • Practice Mode (Jim)
      • Learn Multi-step strategy (Martin)
      • More MRE control for efficiency (not simply forward chain)
      • MRE-ML
  • Phil
    • Explanation interface for system process
    • Software develop cycle
    • build a new data storage
      • for communication and blackboard
      • A mixture of RDF and XML database (for efficient and effective representation of shape etc)
      • Improve performance by avoiding single-threaded program, and single point failure

  • Our position
    • Ready to help Phil and Martin
    • Help formalize "semantic logging" and "explanation template and summary" for transactions and

decisions - overlap with E2ESA

    • Help register special RDF template for background or shared knowledge such as "constraint"


  • Still working on s9 transaction log
  • Today's TAMI meeting canceled, and no discussion/feedback on the proposed data model from other team members (maybe a ppt is too long to be answered)
Personal tools