Re: A proposed provenance wg draft charter

Hi,
Thank you Paul and Luc for putting together this document for discussion.


I strongly agree with what Deborah has pointed out that the charter for the working group should take into account as many existing provenance models as possible and select the most relevant set of provenance terms towards a standardization effort.


Luc said: - what is the bias in OPM that makes it difficult to shoehorn PML concepts (ditto for other provenance languages). The mapping exercise, to my knowledge, did not identify any.


Multiple issues were indeed identified during the mapping process, I am specifically referring to Provenir ontology to OPM mapping where Luc agreed with me that many of the concepts in Provenir represent broader notions of provenance as compared to OPM (provenir:data vs. opm:artifact) . 


Further, many fundamental provenance related, non-causal properties that are modeled in the Provenir ontology are not present in OPM. E.g. provenir:adjacent_to, provenir: transformation_of etc. 


As part of the mapping report work, we also identified many provenance terms that are clearly not present and given the current structure (as defined in the OPM spec v1.1) would not be possible to represent.


There are many specific issues in OPM related to the inferencing, modeling of properties etc., which I have repeatedly pointed out to Luc, Paul and Paolo and given specific examples from the Provenir ontology. 


Thanks.


Best,
Satya



----- Original Message -----
From: Luc Moreau <L.Moreau@ecs.soton.ac.uk>
Date: Friday, October 15, 2010 9:55 am
Subject: Re: A proposed provenance wg draft charter
To: Deborah McGuinness <dlm@cs.rpi.edu>
Cc: Paul Groth <pgroth@gmail.com>, "public-xg-prov@w3.org" <public-xg-prov@w3.org>, paulo <paulo@utep.edu>

> Deborah,
> 
> I think we should not forget the work that the incubator has done.
> I do not understand your statement "there is strong a bias when 
> one 
> starts with one
> and tries to shoehorn in the others."
> 
> Let's be concrete, I suppose you are talking about PML,
> - what is the bias in OPM that makes it difficult to shoehorn 
> PML concepts
> 
> (ditto for other provenance languages).
> 
> The mapping exercise, to my knowledge, did not identify 
> any.  Please 
> correct me if
> I am wrong.
> 
> This issue needs to be nailed down. I am happy to have a 
> teleconference 
> at some other
> mutually convenient time. But, we need to be able to justify our 
> position.  It's important
> to do it soon, because we can't write a final report properly 
> with 
> knowing what the conclusions
> are.
> 
> Regarding the "fast" aspect, I think it is crucial too. 
> Otherwise, 
> somebody else
> will do it, de-facto way!
> 
> Thanks,
> Luc
> 
> 
> 
> On 10/15/2010 02:33 PM, Deborah McGuinness wrote:
> >  On 10/15/2010 8:53 AM, Paul Groth wrote:
> >> Hi Deborah,
> >>
> >> Thanks for your response. This is exactly the sort of 
> discussion we 
> >> wanted to start.
> >>
> >> Luc and I wrote the proposal with the notion that we wanted a 
> fast 
> >> working group.
> > Fast working groups are good but I think we should focus on 
> the 
> > outcome rather than fast.  it is not that fast precludes 
> quality but 
> > we should not focus on fast and have that preclude other 
> options that 
> > may be considered more representative of the broader 
> provenance 
> > community.
> >> With that in mind we thought it would be good to start with 
> something 
> >> already existing and not develop a whole new model. 
> Additionally, we 
> >> think that OPM is already a subset of what's out there 
> already 
> >> including PML. I think the DAML+OIL comparison is not so apt 
> because 
> >> OPM was already developed through a community process. 
> > Actually I think the DAML+OIL model is quite apt.
> > DAML was developed by a large DARPA program over years with 
> many 
> > participants.
> > OIL also was developed by a consortium of people as well over 
> time 
> > (and actually there was a little bit of overlap in the authors 
> but it 
> > was relatively small).
> > Both grew up at about the same time.
> > While this is looking back, I would venture to say that if 
> either went 
> > in as a proposed standard without engaging the other 
> community, either 
> > would have had a VERY hard time getting through a standards 
> body 
> > DESPITE the fact that each had a significant community who had 
> worked 
> > on them.
> > What happened was that  an adhoc US/UK working group self 
> formed to 
> > put something together that reflected what the authors thought 
> > captured the essence of both.
> >
> > OPM was emerging in approximately the same timeframe as some 
> of the 
> > other contender provenance models.  I agree of course 
> that it came out 
> > of a community but it did not include much participation from 
> some 
> > other communities.
> >
> > I think we need to look at what is out there now and determine 
> what 
> > the starting point should be and I do not think that OPM 
> reflects the 
> > starting point for some of the other communities.
> >> Furthermore, as we say in the draft, OPM is the starting point
> > My objection is to taking OPM  as the starting point.
> > I think there is too strong a bias when one starts with one 
> and tries 
> > to shoehorn in the others.
> >
> > Paulo in particular had raised a number of issues when he was 
> trying 
> > to look more deeply at connections and embedded assumptions.
> >
> >> and would change over the coarse of the working group.
> >>
> >> Anyway, since you have a lot of experience on these working 
> groups, I 
> >> was wondering how you think we could design the charter to 
> make 
> >> production of a recommendation as fast as possible with the 
> approach 
> >> you have in mind?
> > I agree that having one starting point and refining may be 
> considered 
> > to be a fast approach  but if one really thinks there may 
> be 
> > significant changes, then that may not be the case.  I 
> think though 
> > that it could easily be interpreted as biasing the outcome 
> towards 
> > that starting point (and i would interpret it that way and 
> would be 
> > unhappy).
> > I think though we need to be open to other starting points so 
> i think 
> > we have to be open to other starting points.
> >
> > I see two routes that could be taken (and of course there are more).
> > 1 - write a charter now that takes starting points as input
> > 2 - follow the owl model and form an adhoc working group whose 
> aim it 
> > is to come up with a joint submission for the starting point.
> >
> >
> >>
> >> Also, at least from my reading, it seems that you would 
> endorse 
> >> having a a working group work for a standard provenance 
> model. Is 
> >> that correct?
> > I  support a working group to come up with a recommended 
> provenance 
> > model.
> > I  do NOT support a working group that takes a single 
> starting point 
> > and aims to refine that one to come up with the recommendation.
> >>
> >> thanks,
> >> Paul
> >>
> >>
> >>
> >>
> >>
> >>
> >> Deborah McGuinness wrote:
> >>>  Greetings,
> >>> Thanks for the work on the draft charter.
> >>> I am sending this quickly because i just read the beginning 
> and have 
> >>> a significant problem with the slant of the charter with the 
> >>> position of
> >>> "The goal of this working group is to refine the Open 
> Provenance Model"
> >>> and thought I should air this immediately upon seeing the draft.
> >>>
> >>> I strongly believe that the charter of a working group that 
> would 
> >>> end up with a provenance recommendation should start with 
> the 
> >>> contender models and take the best aspects from many of them 
> RATHER 
> >>> than starting with one model and refining it.  I 
> strongly oppose the 
> >>> position that the charter should take any single model and 
> work to 
> >>> refine it.  I would propose rather that this group 
> would work like 
> >>> the recent RIF working group or others where contender 
> starting 
> >>> points were submitted  or like the OWL working group 
> where two 
> >>> contenders emerged - DAML and OIL and members from each side 
> worked 
> >>> to create a submission - DAML+OIL that truly took elements 
> from both 
> >>> languages and ended up with a joint proposal rather than 
> this way 
> >>> that just takes a single model as a starting point.
> >>>
> >>> There was some disappointment in the group when we chose to 
> map the 
> >>> other contender models to a single model, but most if not 
> all 
> >>> understood that decision from the perspective of time.  
> I can not 
> >>> say though that i (or i expect others who were disappointed) 
> could 
> >>> go along with this position in a proposed charter.
> >>>
> >>> I am sorry that this comes from someone who did not 
> participate 
> >>> nearly as much as I had hoped to.
> >>> I had an extremely bad year with my mother's multiple 
> >>> hospitalizations and general overwhelming health problems 
> and her 
> >>> subsequent passing.   I simply have had no choice 
> but to put 
> >>> everything other than family health at a much lower priority.
> >>>
> >>> Deborah
> >>>
> >>>
> >>>
> >>>
> >>> On 10/15/2010 7:58 AM, Paul Groth wrote:
> >>>> Hi All,
> >>>>
> >>>> Today on the call we are scheduled to talk about 
> preparations for 
> >>>> the final report. Luc and I feel that to write a compelling 
> final 
> >>>> report we should be clear about exactly what the report 
> should 
> >>>> recommend. There has been some consensus that a working 
> group 
> >>>> should be formed around the recommendations extracted from 
> the 
> >>>> scenarios ( 
> >>>> 
> http://www.w3.org/2005/Incubator/prov/wiki/Recommendations_for_scenarios). 
> >>>>
> >>>>
> >>>> To that end, we have prepared a draft working group charter 
> ( 
> >>>> http://users.ecs.soton.ac.uk/lavm/draft-charter.html ). We 
> note 
> >>>> this is only *our own* proposal and we see this as a 
> starting point 
> >>>> for discussion within the group.
> >>>>
> >>>> We look forward to any comments, questions, thoughts about 
> this 
> >>>> proposal. We hope this helps the group to continue to 
> coalesce 
> >>>> around a way forward.
> >>>>
> >>>> Thanks,
> >>>> Paul and Luc
> >>>>
> >>>>
> >>>>
> >>>
> >
> >
> 
> -- 
> 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
> 
>