wiki:use-cases/create-qc-record

Use-Case: Modify or Create a quality control record


Author: Bryan Lawrence
Version:V 0.1
Date: 27 July 2010
Associated Work Packages:

  • WP2 - CIM definition
  • WP4 - Services
  • WP6 - CIM creation

Purpose: Quality control records will be created for data and metadata.

Actors: RegisteredUser, MetadataManager, DataManager, CIMToolkit

Related Use-Cases:

Annotate a CIM record

Summary:

Description of the key activities which are expected to occur associated with quality control (for ESG as an initial use case, but this is relatively generic).

Typical Course of Events

There are number of sub use cases.

  1. Data is automatically produced, and some automatic tool (part of the CIMToolkit) produces some sort of quality control checklist which should be recorded.
    • CIM records pointing at that data should be associated with the quality control information, and
    • services showing these records should show the existence of this information.
  2. A manual investigation of the data shows either some issues with, or makes a statement about the quality of, data. A DataManager wishes to record this information.
    • CIM records pointing at that data should be associated with the quality control information, and
    • services showing these records should show the existence of this information.
  3. An automatic tool (part of the CIMtoolkit) parses existing metadata and finds some issues (e.g. broken links).
    • The relevant CIM metadata records should be associated with the quality control information, and
    • services showing these records should show the existence of this information.
  4. A manual investigation of the metadata shows either some issues with, or makes a statement about the quality of, metadata. A MetaDataManager wishes to record those issues.
    • The relevant CIM metadata records should be associated with the quality control information, and
    • services showing these records should show the existence of this information.
  5. A formal citation has been produced, and the records should be updated to show this information by a MetadataManager.
    • CIM records should be updated to show their formal citation.
  6. An issue is resolved.
    • Records should be modified to show issue has been resolved.
    • User interfaces may wish to hide (but not delete) the fact that there was an issue.

Alternative Course of Events

Important to note that in a distributed environment, the qc observations could be made in multiple locations, corresponding to multiple copies of an original document instance - none of which may be the originator!

Pre-Conditions

CIM records exist for data and metadata entities.

Post-Conditions

Either the CIM records have been updated, or new ones created. Either way, they need to be exposed for harvest.

Miscellaneous Notes:

Key process for interacting with ESG for CMIP5.

Prompted by the observation that currently (July 27, 2010) the CIM quality control class is expected to be embedded into anything carrying the document stereotype. I have a problem with that in that it doesn't fit the expected workflow:

  1. Someone creates a CIM document
  2. Someone (possibly someone else) makes q.c observations
  3. It's currently not obvious to me how one makes observations about the data as opposed to observations about the metadata.
  4. It worries me that the documents should not be changed when someone makes a qc observation, since it's about the document. I think q.c. should (mostly) be something that comes from a query - tell me what q.c. documents apply to this record - which obviously has interface issues.

The Q.C. class looks like this (from trunk): source:CIM/trunk/quality/quality.png

Use-Case Diagram: