ProvenanceAnnotation Session

1
Provenance/Annotation Session
Workshop on Data Derivation and Provenance
Chicago, October 17-18, 2002

• Jim Myers, PNNL
• Sanjeev Khanna, U. Penn.
• Peter Fox, UCAR

2
Session Outline

• Overview/Issues Jim Myers
• Examples and Categorization of Provenance/Annotati
  on types Dave Pearson, Oracle UK
• Complexity of Tracking P/A
• Wang-Chiew Tan, UCSC
• Facilitated Discussion Peter Fox

3
What is Provenance/Annotation?

• Dimensions (1)
• Personal/group ? public/published scope
• Many ? Few
• Informal ? Formal
• high value to few ? low value to many
• Conjecture ? Conclusion

• Ask Dave (next talk)

4
What is Provenance/Annotation?

• Dimensions(2)
• Associated with files, parts of files, parts of
  multiple files, DB entries, DB Query Results,
• Human use ? Machine Use
• Aid discovery, an alternative to storage, a
  record/audit trail, an alternative to legacy
  papers

5
The P/A Lifecycle
Community
Personal/Group
6
Issues

• Common Terminology
• Types of P/A
• Useful Categorizations
• Uses of P/A

7
Issues 2

• Variations in
• Cost/Benefit
• Who pays, who gains
• Delay between costs and benefits

8
Issues 3

• Write many, read one
• Alternative encodings of P/A (directory
  structures, tacit knowledge, hardcoded
  assumptions,)
• Schema Evolution

9
Questions

• To what extent do P/A issues/strategies differ by
  category, community, use?
• At what scale does coordination/standardization
  makes sense?
• How much of lifecycle can be automated?
• How can researchers participate in multiple P/A
  cycles
• What uses have proven benefits?
• What economic models have been shown to work?

10
Next talks

• Dave Pearson Examples and categories of P/A
• Wang-Chiew Tan Discussion of complexity of
  managing P/A data versus categories