This is to track the discussion on the most appropriate solutions for the historical issue of additional, non-defining relationships such as "part-of" relationships.

Briefing Paper for the discussion on additional, non-defining Relationships                                      16th November 2015

Purpose

To document the discussion and proposed strategy for the maintenance of additional, non-defining relationships.  This currently consists entirely of, but is not necessarily restricted to, “part-of” relationships.

Background

• In 2005 the decision was taken to discontinue maintenance of defining part-of relationships until broader issues in the anatomy model were resolved.  Rather than retiring all part-of relationships they were made “additional” (ie) excluded for the purposes of classification.

• Additional Relationships were removed from the Stated Relationship files so that they would not be passed to classifiers, since classifiers cannot use non-defining relationships.   

• The only maintenance that is now required on these relationships is that they be inactivated whenever either the source or destination concept is inactivated - in order to avoid publishing an invalid active relationship.   This inactivation is performed automatically by authoring tools.

• The problem with this approach is that these additional relationships are created and maintained by Authors and as such are being ‘Stated’.   Normally we would expect authors to make changes which appear in the stated relationship file, which is then classified to produce the inferred relationship files.   When part-of relationships appear only in the Inferred File, we have an ‘effect’ with no apparent ‘cause’.

Proposal

The issue has been discussed with subject-matter experts, who agree that the most effective solution would be to insert the additional relationships into the Stated relationship files, which would then be used to create the Inferred relationships files as per the normal process. This would be actioned through the Delta files, with a current effectiveTime, in order to correctly preserve the historical audit trail.

This will mean that parties who currently classify stated relationship files will need to ensure that their process removes any non-defining relationships prior to classification.

As the potential impact here is significant, the proposed timeline for implementation should be in line with the July 2016 International release, as opposed to the January 2016 release.

This approach needs to be socialized and approved by all parties.

Next Steps