Re: prov-o review / comments

On Feb 21, 2012, at 12:15 PM, Paolo Missier wrote:

> Hi,
>  as requested, a few comments on the ontology (as of now: it's a rapidly moving or perhaps crystallizing target so some of the comments may have been superseded already)
> 
> ==== A) on class hierarchies:
> 
> 1 ====
> dm says: hasOriginalSource is a strict sub-relation of wasDerivedFrom.
> but in -o it's a sub-property of wasAssociatedWith

currently:

hadOriginalSource subprop of wasDerivedFrom
wasDerivedFrom subprop of tracedTo
tracedTo subprop of involved.

So #1 has been addressed?


> 
> 2 ===
> tracedTo property hierarchy. some subclassing is part of DM, notably
>  wasDerivedFrom  implies tracedTo
> but I am not sure about others, e.g. specializationOf,

specOf is a top level property now.

> wasAttributedTo,

I added an annotation into the ontology: "TODO: Paolo questions this subproperty axiom."


> and more.

I'll need explicit names to  hunt them down.

> 
> Is there a justification for this hierarchy?


If not, there can be, using annotations within protege.
https://www.w3.org/2011/prov/track/issues/267


> 
> ====  B) on inferences:
> 
> Some inferences are captured, namely those that map to subclass relations, while others are not. Was this done systematically? I didn't check throughout  but for instance
> 
> wasQuotedFrom =>  wasAttributedTo in DM, but not in -o.

I moved wasQuotedFrom to be a subproperty of wasAttributed
> wasQuotedFrom =>  wasDerivedFrom in DM, but not in -o.
> 

We can't point to both an Activity and an Entity :-/

:snippet prov:wasQuotedFrom :novel

but I don't want to say :snippet prov:wasAttributedTo :novel . 




> there is a rdfs:comment on this though:
> "TODO: Shouldn't Quotation be a subtype of Derivation (and same for the binary relations?) -Tim  -- "

So, should it?


> 
> ==== C) what is the intended usage of the involvement property (not the Involvement class)?

What about the comment is not clear or needs to be elaborated?

> 
> ==== D) is the *qualified* property still needed?

No, I renamed it to prov:involved.

> 
> ==== E) 6.6 wasSummaryOf is a strict sub-relation of wasDerivedFrom.
>  this is not the case in -o
> 

Moved and added annotation justifying it (citing you).

> (then again, summary may disappear in the future)
> 
> ===  F) Trace Class
> 
> rdfs:comment says "A prov:Trace can be from any prov:Element to any prov:Element, so it cannot be a subclass of prov:EntityInvolvement or prov:ActivityInvolvement."
> 
> but then Trace is in fact a subclass of EntityInvolvement?


I don't understand. Whose "in fact"? DM?


> 
> ==  G) equivalent classes EntityInvolvement (asnd  ActivityInvolvement)
> 
> why not just subclasses of (entity some Entity)? note that an OWL reasoner won't do anything with these equivalences at the moment.

The equiv class stuff was my fault. Protege hides to much for this turtle-typer.

> I think Stian recently addressed this


He did, and then yanked them away due to ISSUE-265




> 
> ==== H) class Role
> 
> is this class still needed?  isn't this subsumed by general attributes?

Generally, it's good to define the class for the range of your property.
I know that http://www.w3.org/ns/sparql-service-description# had to grow that way during last call.

> and if we keep it, currently the domain of hadRole includes Derivation, however this seems incorrect as there are no roles associated to derivation

I changed domain of hadRole to Roled.
I added Roled as a superclass of many non-Derivation Involvements.

http://dvcs.w3.org/hg/prov/diff/374e16567121/ontology/ProvenanceOntology.owl





> 
> ===
> Collections missing (Stian working on this I think)

Yup.

-Tim


> 
> 
> --Paolo
> 
> 
> -- 
> -----------  ~oo~  --------------
> Paolo Missier - Paolo.Missier@newcastle.ac.uk, pmissier@acm.org
> School of Computing Science, Newcastle University,  UK
> http://www.cs.ncl.ac.uk/people/Paolo.Missier
> 
> 
>

Received on Friday, 24 February 2012 06:23:04 UTC