Re: PROV-ISSUE-8 (define-generation): Definition for Concept 'Generation' [Provenance Terminology]

Jun Zhao wrote:
>> When I looked at both OPM and PML, I thought there was a core of ideas 
>> common to
>> both that included "resources", "processes", "provenance" and some 
>> fundamental
>> ways in which they are related.  I think it's this core of inescapable 
>> notions
>> and their relationships that's maybe more interesting than their 
>> individual
>> definitions.
> 
> Would you care to clarify this?

OK, let's take this illustration from the OPM vocab spec 
(http://open-biomed.sourceforge.net/opmv/ns.html) as an example:

http://open-biomed.sourceforge.net/opmv/img/opmv_main_classes_properties_3.png

When I view this image, I see three key concepts:

   "artifact"
   "process"
   "agent"

But what really makes them work together as elements of a provenance model is 
the relationships between them:  "used", "wasGeneratedBy", "wasControlledBy", 
"wasDerivedFrom".

And when I look at PML, I find, for example at 
http://tw.rpi.edu/portal/Proof_Markup_Language#Vocabulary, a diagram containing:

   "NodeSet"
   "InferenceStep"

related by "hasAntecedent" and "isConsequentOf".

To my mind, at a high level of abstraction, "NodeSet" is a case of "artifact"; 
"InferenceStep"is an instance of "process", "hasAntecedent" is an instance of 
"used", isConsequentOf" an instance of "wasGeneratedFrom".

Because PML is particularly focused on formal proof structures, there are  many 
additional concepts there that don't have a corresponding explicit recognition 
in OPM, but at this stage I think it's the high-level, general purpose concepts 
we need to isolate.

...

Interestingly, in attempting to answer this question, I failed to locate the 
source of information about PML that I had read previously, where I thought 
there was a more complete summary.  I wasd looking for a visualization of the 
OWL provenance module ontology for PML, but a quick search failed to find one. 
(As a suggestion to the RPI people here, an easily accessible E-R style diagram 
of the elements of your ontologies, module-by-module, might make your materials 
easier to absorb.)

One thing I did find, which I read as supoporting my position here, is the RPI 
submission for the 3rd OPM challenge, described here:

   http://twiki.ipaw.info/bin/view/Challenge/TetherlessPC3

which lists the following mapping of concepts, via an RDF vocabulary called 
"ProtoProv":

ProtoProv?              OPM                 PML
----------              ---                 ---
ProtoProv?:Variable     Artifact            pmlj:NodeSet
ProtoProv?:Function     Process	            pmlp:InferenceRule
ProtoProv?:Controller	Agent	            pmlp:InferenceEngine
ProtoProv?:Usd	        Used	            pmlj:hasAntecedentList
ProtoProv?:Wgb	        WasGeneratedBy?	    pmlj:isConsequentOf
ProtoProv?:Wcb	        WasControlledBy?    pmlj:hasInferenceEngine

I also found reference to a comparison of OPM and PML on the RPI Tetherless 
World Constellation wiki, but the document itself is not available for public 
download (http://tw.rpi.edu/wiki/Image:TW-2009-21.pdf, via 
http://tw.rpi.edu/wiki/A_Comparison_of_the_OPM_and_PML_Provenance_Models).

#g
--

Received on Thursday, 2 June 2011 14:46:42 UTC