Re: prov-wg Implementation Report Review (Due Fri Feb 15)

Hi Paul, Dong, Stephan,



Thanks for producing the document. It looks very good, and summarise
results neatly.

A few comments/suggestions:

1. What is the plan for this document? Is it to make a note of it? It 
would be nice for it to have
some formal status fot it.  If note, should it be part of the PROV 
family of documents?

2. abstract: I don't think the first sentence is right. We don't report 
on implementation and usage of
the prov family, but only the 4 normative specifications of the prov 
family of documents.

-> This document reports on implementations an dusage of the four
    normative specifications [prov-o, prov-dm, prov-n,
    prov-constraints] of the PROV Family of Documents [PROV-OVERVIEW].

3. section 1: bullet 2: "We document that the prov-constraints
specification is implementable". The intro was saying you were taking
two approaches.  So what is the approach?

4. section1: bullet 1:
    we document that there are multiple ... -> we document the existence 
of multiple ...
    we document that there are at least ... -> we document the existence 
of at least ...


5. Section 1.1. I would make it section 5, after the facts have been 
presented

6. Section 1.1. I would not try to paraphrase the exit criteria. The 
risk is that the new wording
    may introduce a different semantics.

7. Section 1.1, I would be factual. To start with just number the Exit 
criteria.

   PROV CR Exit Criteria 1.a PROV-O
    ...
   PROV CR Exit Criteria 1.b PROV-O
    ...
   PROV CR Exit Criteria 2.a PROV-N
    ...
   PROV CR Exit Criteria 2.b PROV-N
    ...
   PROV CR Exit Criteria 3 PROV-DM
    ...
   PROV CR Exit Criteria 4 PROV-Constraints
    ...

8. Section 3.1 shows that a minimum of 4 implementations produce and 
consume all constructs defined in prov-dm.
    Do you mean BOTH produce AND consume,   or EITHER produce OR consume?

    This said, it's not a justification for us meeting PROV-DM exit 
criteria.
    We should say:     Met because PROV CR Exit Criteria 1.a, 1.b, 2.a, 
2.b were all met.


9. I don't think "For PROV-O, the implementations are from three 
separate institutions" is precise enough.
    It could be satisfied by one pair being from one institution.
    I would be more factual (reusing your summary):

    PROV CR Exit Criteria 1.b PROV-O is met because:

    The ProvValidator (University of Southampton) validates (Consumes) 
all PROV-O terms generated by PROVoKing (King's College London)
    The prov-check (VU University of Amsterdam) validates all PROV-O 
terms converted by the ProvToolbox (University of Southampton)

10 For prov-n, I would also reuse your summary.

      PROV CR Exit Criteria 2.b PROV-N is met because:
     - The ProvValidator (University of Southampton) validates 
(consumes) all PROV-N terms generated by the Provenance Server and the 
PROV-Python library (University of Southampton, Python code base)
     - The ProvValidator (University of Southampton) validates 
(consumes)  some PROV-N terms generated by APROVeD (Ghent University)
     - The ProvValidator (University of Southampton) validates 
(consumes)  all PROV-N terms from the examples in the PROV-DM document 
(Provenance Working Group)

PROV-DM documents -> PROV-DM document



11. It would be nice to say something along the lines of "The WG 
recognizes that implementing the PROV-Constraints document requires 
substantial effort. It is nice to see that three radically different 
appraoches were chosen to implement this specification: SPARQL, Java, 
Prolog, which speaks for the implementability of this specification."


12. Prov-json: check capitalization.




On 11/02/13 21:41, Paul Groth wrote:
> Hi All,
>
> You can find a draft of our implementation report at:
>
> https://dvcs.w3.org/hg/prov/raw-file/default/reports/prov-implementations.html
>
> A couple of notes:
> - Please let us know what you think.
> - Comments due by Fri Feb 15 so that we can process the comments in-time
> - Thanks to Tim & Stephan for their usability comments.
> - We will run the scripts again next week. So please if you have any 
> more implementations or datasets that use PROV, or know of anybody who 
> does, tell them there's still time to fill one of the surveys in. 
> Given that we will add the acknowledgements next week.
>
> Finally, thanks to Dong who did a brilliant job of generating the 
> tables within the report.
>
> Thanks
> Paul
>
> P.S. We now have more reported implementations (at time of report) 
> than SKOS, OWL2, SPARQL, RIF, RDFa, and RDF :-)
>
>

-- 
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