A Node Set has a self-referential relationship “relates to”. In GSIM 1.1, Statistical Classification had a self-referential relation “is based on”. So, in fact through inheritance, Statistical Classification had two self-referential relationship – “relates to” and “is based on”. This is a little redundant. In GSIM 1.5, the “is based on” relationship has been changed to “is variant of” and the “Variant” and “Variants Available” attributes were removed. What do you think of this change?

Feedback from countries

CountryResponse in shortFeedback from country
CroatiaOK with the changeThe Croatian Bureau of Statistics supports the change.
LithuaniaOK with the changeConcerning change of the “is based on” relation, in our practice we most commonly use this relation also in other domains of our work in Statistics. We have no objection to “is variant of”, although in our understanding scope of the term is a little bit broader. In our understanding “is based on” permits smaller change than “is variant of”. But it might be just our semantical habit. The change permits elimination of the redundancy which seems reasonable.
FinlandOK with the change

After careful consideration, we are happy with the new solution. "Is variant of" is a special kind of relationship between Statistical Classifications. The other relationships they inherit from the Node Set.

Variant and Variants Available -attributes for Statistical Classifications can be excluded as this information can be derived from the Is Variant Of -relationship.

MexicoOK with the change

We agree. The change sets a way to describe a new statistical classification as a variation of another. It provides a better mechanism to document classification versions. And the “Predecessor” attribute would have the same use than “Variant”. Hence, we consider eliminating “Variant” correct. “Variants Available” would be relevant for managing purposes, as this information could be used to control various versions of a Statistical Classification.

AustraliaSo OK...?

See comment from ABS on issue #50 from 16 July 2018.

Basically the proposed change makes sense at the conceptual level.

The post outlines a couple of implementation considerations that don’t preclude the conceptual “tidying” of the model, including backwards compatibility. Those who have already implemented GSIM 1.1 may choose not to update their existing implementation due to change management costs because the marginally “cleaner” approach in GSIM 1.5 does not deliver sufficient value to redo implementation.

CanadaOK with the changeWe agree with the change, Classification relates to itself through variant. The relationship is very clear.


  • No labels

1 Comment

  1. InKyung Choi

    26 Oct, 2018

    (GSIM Revision Meeting 24th October, 2018)

    Keep as changed