W3C home > Mailing lists > Public > public-xg-prov@w3.org > October 2010

RE: Overview presentation

From: James Myers <jim.myers@computer.org>
Date: Fri, 29 Oct 2010 17:08:02 -0700
Message-ID: <BAY138-W2113280C8DCCD75CBFE768BF460@phx.gbl>
To: <paulo@utep.edu>, <gil@isi.edu>
CC: <public-xg-prov@w3.org>

Paulo,
You've done interesting things with PML, but what feature(s) of the PML language make it more suited to doing what you did that the other languages being discussed in the group? That, and not what has been done with PML, is what we need to appreciate as a spec group. It seems like OPM or any other language could be served over the web just as easily, and I don't see any problem in documenting the direct provenance of a query ("directions from A to B") that was sent to a web service that then returns an incorrect route. (In the later case, I suspect you're after more - you want to record that a person wanted to get from A to B and therefore submitted a query to get directions  - documenting the 'why' motivation for the web/computational processing that occurred. If you think that is useful for the group to consider, it would be useful to know what construct in PML enables that and hear your opinion on whether that construct has proven itself useful enough and straight forward enough to add to a standard at this point or if further work is needed there.)
 
  Jim
 
 
> Date: Fri, 29 Oct 2010 11:34:17 -0600
> From: paulo@utep.edu
> To: gil@isi.edu
> CC: public-xg-prov@w3.org
> Subject: Re: Overview presentation
> 
> Hi All,
> 
> This alignment with W3C architecture is the missing point for the group 
> to better appreciate what PML has accomplished in terms of provenance. 
> For instance, we already have PML services wrapping key scientific 
> repositories in the US (the USGS gravity and magnetic database, 
> NSF-funded IRIS repository of seismic data, NCAR CHIP processes) and 
> scientists are capable of retrieving either data or data and their 
> provenance in form of PML. Please note that it was possible to implement 
> these services because we spent a long time understanding these 
> databases were put together, which kind of processing the raw data went 
> through and so on.
> 
> Another interesting result of this W3C architecture alignment is the 
> introduction of the notion of a question/answering context. For 
> instance, in PML, we often assume provenance to be the provenance of a 
> resource and the resource to be the outcome of some question that was 
> asked. For example, a person may want to know the provenance of a given 
> route provided by a GPS system because the route appears to be 
> incorrect. In PML, we consider to be important to encode the fact that 
> the allegedly incorrect route was generated because this person was 
> asking for directions between points A and B.
> 
> In the example above, for instance, it may be hard or impossible to 
> infer from an OPM provenance what was the original motivation for the 
> system to generate such route and more specifically to know that the 
> original goal was to get an accurate direction between points A and B. 
> Moreover, when it comes to workflow systems, we see that workflows 
> produce results that are not necessarily the final answer that the 
> scientist have in mind at the time the scientist asked for workflow 
> execution. In fact, the knowledge of this original motivation may be key 
> for one to see what is going wrong with the derivation of a scientific 
> product.
> 
> Many thanks,
> Paulo.
> 
> On 10/29/2010 11:21 AM, Yolanda Gil wrote:
> > All,
> >
> > A few months back I mentioned it would be good to have a slide
> > presentation of the activities in our group. I have been collecting
> > materials over the last few months, and I have posted such a
> > presentation at:
> >
> > http://www.w3.org/2005/Incubator/prov/wiki/File:Provenance-XG-Overview.pdf
> >
> > Feel free to use it, with proper credit to the group.
> >
> > I will update it once more before we finalize our activities.
> >
> > Yolanda
> >
> >
> >
> >
> >
> 
> 
 		 	   		  
Received on Saturday, 30 October 2010 15:50:25 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Saturday, 30 October 2010 15:50:27 GMT