Provenance meeting with Norman on 05/03/2008

Minutes

• T1 model
  • lack of a formal description

• OPM model
  • import / export
  • How do our requirements map to the OPM
    • What does the OPM support?
    • What does it lack?

• Khalid's document has already been distributed.
• Paolo's document will be circulated between us later this week.
• Stuart and Ians documents need aligning.
• Need to characterise queries.
• Stuart wishes to split the 3rd use case (incremental and interactive debugging) into 2 use cases - the first being a simpler version of the current use case that only supports running a worfklow to completion.
• Debugging use case 5 (correct/incorrect usage of a processor) highlights the need to query over multiple provenance executions. This raises questions of how provenance is stored (1 store instance for each run, or all runs in a single store).
• In most use cases the queries need to be interactive - this creates particular performance constraints.

• All documents will be circulated together on Friday. Rather than as attachments to a single email, we will gather them in a wiki page for circulation.
• Need to encourage feedback.
• Stuart will physically hand Carole the documents.

• Need to start prototyping:
  • what are the particular queries?
  • document:
    • example queries
    • our limitations
    • problems?

• Security? Our final document needs to provide a statement on security assumptions.
  • Strawman Proposal:
    • (1) Each workflow writes provenance to a provenance store set by the person running the workflow.
    • (2) The person running the workflow must have permission to write to that store.
    • (3) The responsibility for permissions at a finer grain that the store is out of scope for Taverna.