IW Meeting 2011-05-05

From Inference Web

Jump to: navigation, search

Contents

Meeting Information

Attendees

  • Jim McCusker
  • Li Ding
  • Yangfan
  • Cynthia
  • Deborah


Deborah - will only be on the phone at the beginning. I am requesting that li or jim m. or cynthia be host (and confirm who will do this). th

Jim - I'll do this, but I need to know if we're doing skype for sure, or calling in via a conference line.


Agenda

  • geoff and paper status
  • move todo items from last week
  • follow from provenance telecon

Geoff Paper Status

Combination is improving, but not seeing any interesting combinations between independent proofs.

Do we have good examples?

  • geoff can twik some examples from AURA problem (SRI biology text book)
  • existing TPTP proofs are not generating good enough examples
    • proofs are so different for combine
    • also examples are "well" constructed so that there is little redundancy (and thus few options for getting answers along different paths)
  • TODO (csc) Run proof combiner on AGATHA.
  • Discussion of additional data from NSF cyber-infrastructure
    • Paulo: "skewing" data for proof combination not useful.
    • Li: looking for a valid combined proof case in science.
    • Paulo: in oil industry, can find locations of oil in multiple ways.
    • Li: Asking if we have this kind of data in PML.
    • Paulo: Has it for one, working on getting it for another.

Outstanding Action Items

Example that breaks Paulo's desired disjointness: retrieve file from web. URL is the Source, File is the Information Grab the 5th row of that File. File is a Source, that Line is the Information.

UTEP visit to NSF

Report from w3 provenance telecon

  • Discussed establishment of the group, whether we agree to the 4 task forces proposed.
  • Discussed invited speakers
  • Paulo worried about the prevelance of OPM terms in examples and discussion
  • Should have a more level starting point.
  • http://tw.rpi.edu/web/doc/towardscmo is a possibility for starting with a common, independent vocab, then building a model.

Notes from Yangfan

  • Assumption that the steps are independent, may not be valid for all provenance from inference web.
  • Discussion of uncertainty propagation.

Cynthia's Visit Last Week

  • Deborah: Very successful.
Personal tools
Navigation