PIL OWL Ontology Meeting 2012-02-13

From Provenance WG Wiki
Revision as of 18:50, 13 February 2012 by Tlebo (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Meeting Information

prov-wg - Modeling Task Force - OWL group telecon

Attendees

  • James
  • Mike
  • Satya
  • Stephan
  • Tim
  • Daniel
  • Deborah
  • Stian

Agenda

1. Mapping between PROV DM TPWD constructs and PROV-O http://www.w3.org/2011/prov/wiki/ProvRDF

2. Update on blog post describing use of PROV-O to model simple provenance scenario

Discussions

Elements

1. Entity http://www.w3.org/2011/prov/wiki/ProvRDF#Entity

Tim: Do we include "Element" as a superclass of Entity? Group agreed to exclude 


2. Activity

TODO: Activity, make parallel qualified and unqualified Time (with timestamp and with Time)

 Activity is a class
 wasStartedAt, wasEndedAt as two object properties
 Time is a class (from OWL Time ontology)
 
 wasStartedAt ""^^xsd:dateTime
 wasStartedAtQualified [ a prov:Time; prov:inXSDDateTime ""^^xsd:dateTime ] .
 
 wasEndedAt ""^^xsd:dateTime 
 wasEndedAtQualified [ a prov:Time; prov:inXSDDateTime ""^^xsd:dateTime ] .

3. Agent Agent as a class and a subclass of Entity. (Todo: update the wiki)

Daniel: Add three subclasses of Agent according TPWD - Person, Organization, SoftwareAgent


4. Note James: Have a Note class using annotation relations, would be fine with using built-in RDFS, OWL properties


Note is a class   Suggestion to DM: How is note class related to class Entity and class Activity?  

Relations

 1. Generation  James: Need to clarify what is optional and what is required?    +1 - DM does not distinguish between optional-inferrred and optional-none     Tim: Only create the corresponding RDF nodes for expressing the given optional parameters. For instance wasGeneratedBy([], e, [], [], attrs) then we need to generate the [id] to attach the attrs. ]

 Daniel: Instead of having a qualifiedGeneration id we could have e qualifiedGeneration id    attributes:  an optional set of attribute-value pairs attrs that describes the modalities of generation of this entity by this activity.

Stian: Not just Generation is an event in DM, also Usage. In F2F2 we mentioned that prov:QualifiedInvolvement could be seen rather as a prov:Event - which is less activity-bound.    Generation as an OWL object property will need both entity and activity id.  If activity id is not available, the associated information (attr-values) can be modeled using Generation (subclass of QualifiedInvolvement)


2. Usage No change from existing PROV-O

James: How to address repeated usage by (same) activity and at different points of time

We use the binary (unqualified) and qualifiedUsage

http://www.w3.org/TR/prov-o/#hadtemporalvalue can be used with any QualifiedInvolvement (but HTML does not say so) --- James: OK, that would fix it.

OWL allowed it:

    <owl:ObjectProperty rdf:about="hadTemporalValue">
        <rdf:type rdf:resource="&owl;IrreflexiveProperty"/>
        <rdfs:label xml:lang="en"
            >has temporal value</rdfs:label>
        <rdfs:range rdf:resource="&time;TemporalEntity"/>
        <rdfs:domain>
            <owl:Class>
                <owl:unionOf rdf:parseType="Collection">
                    <rdf:Description rdf:about="Activity"/>
                    <rdf:Description rdf:about="Entity"/>
                    <rdf:Description rdf:about="QualifiedInvolvement"/>
                </owl:unionOf>
            </owl:Class>
        </rdfs:domain>
    </owl:ObjectProperty>
   


3. wasAssociatedWith Suggestion to DM: Agent should be associated with Activity (reverse direction of relation) or rename the property Stian: Responsibility in context of Agent also corresponds to Agent associated with Activity   a) wasStartedBy Ok, we agree on the representation (not to be considered from F2F?)  (collections?)


(Stian: I think the thoughts were something like: if prov:Start is a prov:Event which has a time stamp, then that is the qualified start of the activity - no matter the agent (which wasStartedBy requires to be present))


b) wasEndedBy Ok, we agree on the representation (not to be considered from F2F?) 



4. Responsibility Ok

Daniel: subordinate and responsible can also be roles Tim: Both of these can be described using QualifiedDelegation

Stian: QualifiedUsage does not have a link back to the activity. Why do we for the Delegation need a link back to agent2 (the subordinate) Only need prov:qualifiedEntity to ag1 (the master). What about the activity? In usage you did not have 2 entities+ an activity. The activity or the entity is the way into the delegation/usage. The QualifiedInvolvement bare does not have enough information - you need the property in as well. (As we did not include the inverse property like prov:usedByActivity - only prov:hadQualifiedUsage) But here we are linking the QE to the Agent, not to the activity, so there is no "linking back". My point was just why agent2 is any higher priority to have the inverse prop compared to the activity. ;-) ahhh. :D



5. Derivation Tim: Hide prov:steps as 3 subclasses of prov:Derivation Stian: +1 

Stian: @Tim +1 



6. AlternateOf



7. SpecializationOf Tim: Just model them blindly in PROV-O for now, unless a simple solution using rdfs:subClassOf is possible by Thursday.


For 6,7, show how these can be modeled using rdfs:SubClassOf 

alternateOf and SpecializationOf will also be added in OWL file

(who added the qualified forms of specializationOf? )



8. Annotation property Should the property be applicable to anything or is it constrained to PROV specific classes?

DM: "An annotation record establishes a link between an identifiable PROV-DM record and a note record referred to by its identifier"

Activity or Entity is associated with Notes?


Stian: While 'record' is in flux in DM, just make the prov:hasAnnotation have domain owl:Thing and don't re



9. Time as PROV-O class



10. Location as PROV-O class