PIL OWL Ontology Meeting (extra) 2011-09-28

Meeting Information

prov-wg - Modeling Task Force - OWL group telecon

• previous meeting
• date: 2011-09-28
• time: am PT, 13:30 noon ET, pm UK
• Skype
• wiki page: http://www.w3.org/2011/prov/wiki/PIL_OWL_Ontology_Meeting_2011-09-28
• titanpad: http://titanpad.com/N6djeWRSdy
• next meeting

http://www.w3.org/2011/09/26-prov-minutes.html

Agenda

• so what happens when I change a namespace definition in one of your not-on-a-computer ProvenanceContainers? -Tim
• Role discussion: what alternative should we choose?
  • OPMO+OPMV vs "Satya's approach"

Attendees

• Jim
• Stian
• Tim
• Paolo
• Satya
• Khalid
• Daniel

Discussion

ProvenanceContainer Paolo: A wrapper for the provenance assertions. It is not an entity. It becomes an entity when we start asserting things about it. Paolo: You don't make every class in the world a subclass of prov:Entity - an instance become an entity the moment you make provenance assertions about it Tim: does that mean any instance of any class also becomes typed to prov:Entity once it starts being described with PROV predicates? Stian: Do we expect ProvenanceContainer to appear in the OWL/RDF serialisation, or is it merely a model-representation of the container (ie the OWL/RDF file) similar to Document in the DOM API? Account Satya: Multiple accounts and assertions about entities with the same entities within the same RDF graph would be against RDF monoticity requirement Paolo/Stian: Each account in each RDF (named) graph - can then have potential mismatch Satya: Definition about Account in conceptual model does not highlight enough that the account assertions are not neccessarily to all true nonmonotonicity Satya is concerned that all RDF "looks like" one big graph. Tim: we need to maintain the distinction between 1) using PROV to push around RDF and 2) using PROV to push around non-RDF (e.g., images and Word Documents) Paolo: "scope" in terms of programming languages.

Role alternatives

OPMO vs Roles Daniel: What if somebody does not know about Role? Jim: How are assumesRole and complementOf related to each other? Tim: if we drop Roles to literals, we lose the ability to describe them. Jim: Roles definitely can't be literals. http://www.w3.org/2011/prov/wiki/File:IMAG0118.jpg Two proposals Jim: pe prov:wasControlledBy AliceAsAuthor. AliceAsAuthor a Author;

   prov:assumedBy Alice.
   

Interim agreement to use roles as modeling construct and identify potential issues with it. Issue with OPMO + OPMV approach modeling n-ary relations