Feedback to the LC of the Organization Ontology from the Prov WG

Dear Organization Ontology Editors,

The PROV WG reviewed the part of the document about extension to PROV. 
Generally speaking we agree with your following extensions:

  - org:originalOrganization as a subproperty of prov:used,
  - org:resultedFrom as a subproperty of prov:wasGeneratedBy
  - org:ChangeEvent as a subclass of prov:Activity

In addition, we would like to draw your attention to the concept of 
derivation [1] in PROV, which refers to a transformation of an entity 
into another, an update of an entity resulting in a new one, or the 
construction of a new entity based on a pre-existing entity. Based on 
some general knowledge, one might expect that a new organization should 
be a derivation of the old one, by some sort of transformation or update 
or simply setting up a new entity. But does this indeed make sense with 
the use cases that you consider?

This is a strong albeit useful assertion, enabling you to trace the 
history of an organization. However, the semantics of the PROV model 
does not let you infer this relationship by the combination of 
generation and usage, i.e.,  from the following triples:

ex:o2  org:resultedFrom          ex:a1
ex:a1  org:originalOrganization  ex:o1

The PROV model regards ex:o1 and ex:o2 as totally unrelated, unless 
their relationship is explicitly stated otherwise. If the definition of 
derivation does fit your use case,  making use of this relationship in 
your ontology will make it much more in line with the upcoming 
provenance recommendation. So it is a matter to have a think about what 
you intend to achieve by using the prov:used and prov:wasGeneratedBy 
properties. If you want to include derivation in your ontology, then we 
make the following to suggestions:

1. We RECOMMEND that ex:o2 prov:wasDerivedFrom ex:o1 be explicitly 
asserted. (or a subproperty in the org: namespace)
2. Alternatively, you could add a property chain
       org:resultedFrom followed by org:originalOrganization => 

The second point is that the PROV model comes with a set of implicit 
semantics constraints (in its constraint document [2]). Although these 
constraints are not reflected in the PROV-O, we expect that a provenance 
validator, compliant with  this document, will validate provenance 
statements on the Web. Therefore, it might be worthwhile to take a pause 
and think whether the intention behind your ontology might lead to any 
violation to relevant constraints.

For example, the generation-precedes-usage constraint [3] requires that 
the event when a new organization was generated must precede the event 
when the same organization was used in the process of generating another 

Finally, we would also like to bring your attention to the concept of 
invalidation. In the PROV data model we say that an entity can have a 
lifetime. And the invalidation is the start of the destruction, 
cessation, or expiry of an existing entity by an activity [4].

If this concept is adopted in the Organization Ontology, then you will 
be able to specify more precisely that the cease-to-exist of the old 
organization when a new organization was built up. Might this be helpful 
to your use cases?

Please be aware that by using this concept, there are several related 
constraints to bear in mind. An example is the 
usage-precedes-invalidation constraint [5], which means that the event 
when an old organization ceased to exist must follow the event when it 
was used in the org:ChangeEvent activity.

We, the WG as a whole, will be happy to help you with any other issue 
related to PROV.

Hope this helps.

Jun, on behalf of the PROV WG


Jun Zhao, PhD
EPSRC Postdoctoral Fellow
Department of Zoology
University of Oxford
Tinbergen Building, South Parks Road
Oxford, OX1 3PS, UK

Received on Sunday, 25 November 2012 09:43:38 UTC