RE: PROV feedback

Hi Paul,
 
Yes - I did find the UML diagrams within the DM spec to be helpful, overall.  In fact, I thought using class associations to model the attributes that are part of each PROV statement was a good way to include that information in the diagram so I mirrored that approach when I created my diagrams.
 
In particular, the examples (e.g., figures 2 and 3) were great and could be interpreted with little reference to the supporting text.  I was a little confused by others (e.g., figures 5-8) and had to reconcile what I interpreted from the diagram with what I read in the text.  Some of this might be due to personal preferences regarding modeling style.
 
The majority of my confusion was resolved when I made the distinction between terms that were used to reference Activities (an actual process) from those that were used to describe information about a relationship (e.g., "generation" does not describe an Activity).  Once that occurred it was much easier to generate a model that looked simimlar to the diagrams in the DM spec.  Personally, I find the way I illustrated relationships and their associated classes to be more intuitive in part because the associated classes mirror the notation for each relationship (all attributes are shown; again, might be a style preference).  This made it easier for me to tie the text to the diagram and mentally separate the "syntax classes" from the "things" in the model.
 
Also note that while I modeled roles as explicit classes, this was done to more clearly define the associated relationship and it is not something that I think needs to be included within PROV-N.  They were important for me to fully understand PROV but others may think the extra information is unnecessary.  I use role classes frequently and, if I adopt PROV to use with my project, will include PROV role classes when I pull elements of PROV into my model.
 
Thanks,
Bob

________________________________

From: pgroth@gmail.com on behalf of Paul Groth
Sent: Wed 7/25/2012 2:29 AM
To: Freimuth, Robert, Ph.D.
Cc: public-prov-comments@w3.org
Subject: Re: PROV feedback



Dear Robert,

Thanks for your extensive review! We have created ISSUE-463 for your
comment which you can find at:

https://www.w3.org/2011/prov/track/issues/463

We will now discuss in the Working Group how to address your comment.
To not swamp you with email, we'll do this on our internal mailing
list and get back to you with either a resolution, an update on
progress or clarification questions.

You can always check where we are at by looking at the issue above.

One quick question I had, was whether the UML diagrams with the
PROV-DM were useful to you? Did you base your diagrams on those?

Thanks,
Paul

On Tue, Jul 24, 2012 at 8:38 PM, Freimuth, Robert, Ph.D.
<Freimuth.Robert@mayo.edu> wrote:
> To the W3C PROV working group:
>
> This email is in response to the call for comment and feedback on the
> current working draft of the PROV model.  I reviewed the PROV documentation
> following a recommendation from Jim McCusker.
>
> I am impressed with the volume of PROV documentation.  I did not have time
> to thoroughly review all of the documents that comprise the PROV spec, but I
> was able to review the Primer, Notation, and Data Model.  I found PROV to be
> relatively complete and well-documented.  Thank you for all of the effort
> the working group has put into this.
>
> I have attached a document that contains my detailed comments and feedback.
> Thanks to Tim Lebo and Jim McCusker for answering some of my early questions
> via email.
>
> As part of my review and learning process, I also created a UML model of
> PROV.  This allowed me to actively understand the model and see the effects
> of change proposals.  I attached a document that summarizes my proposed
> changes and a series of images from the model that I created (which
> incorporates the proposed changes), which I hope will help make the proposed
> changes more clear to the work group.  This is done in the spirit of
> offering ideas and solutions, not just complaints.
>
> I would be happy to discuss these documents with the work group if my
> comments are unclear (which is likely).  I am interested in your response to
> these suggestions and I look forward to seeing the final PROV spec.
>
> Finally, some context for my interest in this work.  I am looking for a
> provenance model to use for pharmacogenomics.  Now that I feel like I have a
> decent understanding of the model I will evaluate it for use in this
> scenario.  I suspect I will need to extend the model to capture more
> information about why changes were made, not just when and how they were
> made.  This may be outside the scope of PROV.
>
> Regards,
>
> Robert R. Freimuth, Ph.D.
> Associate Consultant
> Division of Biomedical Statistics and Informatics
> Assistant Professor of Medical Informatics
> _______________________________
> Mayo Clinic
> 200 First Street SW
> Rochester, MN 55905
> www.mayoclinic.org
>
> <<PROV_Feedback.zip>>



--
--
Dr. Paul Groth (p.t.groth@vu.nl)
http://www.few.vu.nl/~pgroth/
Assistant Professor
- Knowledge Representation & Reasoning Group |
  Artificial Intelligence Section | Department of Computer Science
- The Network Institute
VU University Amsterdam