Skip to end of metadata
Go to start of metadata

Page for provenance-specific requirements.

(work in progress) First go on the extraction of Provenance Requirements of a particular Golden Exemplar in Astronomy (Quantities Propagation), also based on the Data Provenance in Astronomy notes.

Top-level dimensions: C - Content | M - Management | U - Use.

  • [C] Original input data is there and accessible, as well as the provenance of that data.
    • "Data Observation" process information has to be accessible as part of the original data.
  • [C] Data transformations (incl. transitory data) have to be tracked.
    • Scripts and mathematic equations in the "Data Reduction" process have to be accessible.
  • [C] Decision points inside the Wf need to be tracked.
    • "Flagging conditions" in an execution need to be tracked.
  • [C?] Justification: Output of a Wf formally linked to to the original data and transformations.
    • Results of the update need to be justified by linking to the original data.
  • [C] Versioning of an RO has to be supported when there are modifications.
    • When updating an RO, backtracking accross versions should be supported.
  • [M] Easy access mechanisms to the provenance information need to be implemented.
    • Tools need to support the inspection of provenance information.
  • [U] Perceived data quality of the external sources should be assessed if desired.
    • As in data from services invoked, HTML pages (different quality depending on the experiment).
  • [U] Quality of Service of the external services should be registered.
    • Previous Wf executions should help identify QoS of external sources.
  • [U] Inspection, analysis, etc. of input, transitory, final data in a Wf execution has to be possible.
    • E.g., "Distance/Corrections/Luminosities errors" checking should be possible.
  • [U] Reproducibility and provability of a scientific experiment should be possible.
    • This should include formal details about the scientific method used.
  • [U] Debugging of a Wf should be helped by provenance information.
    • Inconsistencies (entry data format, scripts, in access to external resources) should be identified.
  • No labels