Re: review of prov-xml

On Feb 26, 2013, at 4:28 AM, Luc Moreau <L.Moreau@ecs.soton.ac.uk> wrote:

> 
> Hi Stephan, Curt, Hook,
> 
> It's a good document. Ideally, there are a few things I would like to see
> fixed before the next release, and if not by then, before final
> release.  I recommend release as a next WD.
> 
> Key issues:
> - It's not clear that examples are schema valid. Many ids do not have
>  a prefix, but no default namespace seems to be defined.
>  Can we check all examples against the latest schema?

The examples from the note should be identical to, or very very similar to, the xml examples in examples/eg-40-xml-examples-by-terms/xml.  All the xml examples in eg-40 currently validate using xmllint.  You can run make in this directory to run the validation tests.

> 
> - I would suggest adding a section 4 to define prov:Document (like
>  there was similar section in prov-n), since it doesn't seem to be
>  defined in the main text.

Sounds reasonable, I will take a stab at it.

> 
> - I suggest updating the mime type application with the feedback
>  received from the ietf mailing list, and submitting the revised
>  version to the ietf mailing list for final check.

Agreed.

> 
> Note that I haven't been able to check the schema compiles naturally
> with JAXB, but you gave us assurance it was the case.

localhost:schema zednik$ xjc prov.xsd 
parsing a schema...
compiling a schema...
org/w3/ns/prov/Activity.java
org/w3/ns/prov/Agent.java
org/w3/ns/prov/Alternate.java
org/w3/ns/prov/Association.java
org/w3/ns/prov/Attribution.java
org/w3/ns/prov/Bundle.java
org/w3/ns/prov/BundleConstructor.java
org/w3/ns/prov/Collection.java
org/w3/ns/prov/Communication.java
org/w3/ns/prov/Delegation.java
org/w3/ns/prov/Derivation.java
org/w3/ns/prov/Dictionary.java
org/w3/ns/prov/DictionaryMembership.java
org/w3/ns/prov/Document.java
org/w3/ns/prov/EmptyCollection.java
org/w3/ns/prov/EmptyDictionary.java
org/w3/ns/prov/End.java
org/w3/ns/prov/Entity.java
org/w3/ns/prov/Generation.java
org/w3/ns/prov/IDRef.java
org/w3/ns/prov/Influence.java
org/w3/ns/prov/Insertion.java
org/w3/ns/prov/InternationalizedString.java
org/w3/ns/prov/Invalidation.java
org/w3/ns/prov/KeyValuePair.java
org/w3/ns/prov/Membership.java
org/w3/ns/prov/Mention.java
org/w3/ns/prov/ObjectFactory.java
org/w3/ns/prov/Organization.java
org/w3/ns/prov/Person.java
org/w3/ns/prov/Plan.java
org/w3/ns/prov/PrimarySource.java
org/w3/ns/prov/Quotation.java
org/w3/ns/prov/Removal.java
org/w3/ns/prov/Revision.java
org/w3/ns/prov/SoftwareAgent.java
org/w3/ns/prov/Specialization.java
org/w3/ns/prov/Start.java
org/w3/ns/prov/Usage.java
org/w3/ns/prov/package-info.java

The results are attached as a compressed archive.

--Stephan

> 
> Cheers,
> Luc
> 
> Detailed feedback:
> 
> Abstract: third sentence does not parse
> 
> Intro. Text copied verbatim from dm. It should be clear for the reader
> that this design comes from DM. Maybe also add some links to DM (for
> references and components).
> 
> 2.1 corresponding to terms defined in the PROV-DM.
> 
> Extend PROV-DM to the PROV data model
> 
> 2.2 salami slice citation missing
> 
> 2.2. It's use ... Its
> 
> 2.3 the PROV-N.  ... PROV-N
> 
> 2.3. As in PROV-N, the attributes can
>     ... "attributes"
>  Generally, do we want to go through the document and check whether this "attribute" or attribute
>  I would suggest that a <dfn>prov-n attribute</dfn> is introduced, and <a>attribute</a> to refer to it.
> 
> 2.3 Wherever an "id" is referenced from a later concept
>   Later??
> 
> 2.4 using A the prov:type PROV WITH and ... delete 'a', with???
> 
> This element can be use. ->. Used
> 
> The following examples shows type.   Which example? use number.
> 
> from the default element type. .   Not sure what is meant here
> 
> 2.6 with [PROV-N] record names
>   Prov-n expression names
> 
> section 3
> "The element prov:wasStartedBy is used to reference a prov:Start from within a prov:Document or prov:BundleConstructor."
> The xml names  do not appear in same font/color. I wonder whether some could be links?
> 
> 3.2 HasPrimarySource. ... HadPrimarySource
> 
> example 23.  Line-management example has no prefix, but no default prefix defined. Generally hav we checked all examples are schema valid.
> I now realise prov:ref="a" suffers from the same problem. This occurs in several xamples.
> 
> 3.4.2 The element prov:bundleContent is used to reference a set of nested provenance statements from within a prov:Document.
> I don't think the term reference here is right.  ... What about
>  " prov:bundleContent is used tho package up a set of prov statements and give them an identity.
> 
> 3.7.1. "The identifier attribute is used to identify instances of PROV types or relations."
> 
> I don't think the phrasing is right. We agreed the identifiers identify entities, or activities, etc.
> The sentence seems to indicate that the Id is used to identify the record.
> 
> More generally, I am becoming unsure that the distinction prov:ref / prov:id is right. I wonder whether we shouldn't go for prov:ref everywhere. The only exception is bundleconstuctor which rely identifies a bundle and therefore needs prov:Id
> 
> 
> Is there a distinction between
> <prov:location xsi:type="xsd:string">(5,5)</prov:location>
> And
> <prov:location>(5,5)</prov:location>
> 
> Section 3.7.3.x
> 
> The element prov:value is used to represent a PROV attribute.
> Which attribute? Doing what? Same for all the others.
> I guess it's "attribute".
> 
> 
> -- 
> Professor Luc Moreau
> Electronics and Computer Science   tel:   +44 23 8059 4487
> University of Southampton          fax:   +44 23 8059 2865
> Southampton SO17 1BJ               email: l.moreau@ecs.soton.ac.uk
> United Kingdom                     http://www.ecs.soton.ac.uk/~lavm
> 
> 
> 
>