IW Meeting 2006 Oct 19

From Inference Web

Jump to: navigation, search


  • Calo video
  • demo at LM ATL
    • demo InfoMaster (optional)
    • demo IW tools; Trust; TAMI content
    • status report on PML2 tools
  • PML2 tools
    • OWL-Individual tool (we can offer our code for bootstraping)
      • currently, LM chooses java-based OO to access/store data, i.e., data is accessed in terms of java objects.
      • next version will require OWL-based APIs to access/store data, i.e., data is accessed in terms of owl individuals.
    • PML2 browser
    • PML2 registry (Mayukh will work on it)
  • WineAgent
    • Restored to old version, seems working
    • we still encounter page failure frequently, may because the serverlet on tomcat is not working properly (Mayukh)
    • We still need to investigate if the returned results are correct (Deb)

lm atl visit Goal: decide what to be brought up to the GILA meeting

  • [important] nail down gila ontologies by walking through gila-example1.owl
    • walkthrough dataflow gila-example1.owl
      • have input psteps from FLA/UMD
      • need input ACO from LM ATL
      • need input preference/constraint from GTRI
      • need confirmation from learners and MRE
    • expose explanation interfaces
      • simplified IRL’s explanation, and that could be extended with more details by ILRs
      • MRE explanation
      • the data in the blackboard is fully connected and can be deemed as explanations
      • [optional] inquire explanation interface/requirements
  • [important] ontology development guideline:
    • creation - naming convention
    • mapping/partition/merging/evolution - documenting revision with explanation, tracking changes and discussions.
  • [semi-offline] next move on OWL/RDF
    • the guideline to next step data access API.
      • To move forward from the temporary java-object approach, we need to offer guidelines (probably interfaces/examples) for the owl/rdf approach. The suggestion need to cover features, pros, cons, architecture (view graph, java class).
      • Currently, we have some general purposed java code for managing PML data, this could be a sample/starting-point for LM ATL to build APIs for the whole group.
    • a wild rdf/owl style problem solving method: an agent manages a triple store that is subscribed by many semantic web services (Learners), the goal is the initial content of the triple store. The agent queries the triple store with the input information of the semantic web services, and call semantic web services once the input are satisfied. The semantic web services will take the input (or maybe get more knowledge) and add more knowledge to the triple store. The process will continue running until the goal is solved or failed.
    • the role of owl-s in the RDF/design. Model all learners using OWLS.
Personal tools