Re: PROV-ISSUE-188: Section 5.2.3 (PROV-DM as on Nov 28) [prov-dm]

I think Satya's point is that, while we shouldn't tie in domain-specific
constructs, we should be supportive of those domains. If there is something
in a generalized model that doesn't work for a particular domain, then it's
not generalized across that domain.

I think that what Satya is looking for is the conjunction of artifact and
agent (ArtificialAgent). This *could* be extended in domain models by
adding an Artifact class and conjoining that and Agent, but I think his
point is that a software agent is also an ArtificialAgent, and it's not
good modeling practice to have to extend a general model from a non-leaf,
for instance, by declaring a superclass for SoftwareAgent.

I would argue that scientific instruments are more than just software
agents. In fact, since we're often talking about embedded software, the
software itself can't work without the specialized hardware. For our
purposes, it's the two together that make the agent, not just the software.
I think this is important for the domain of science provenance, which I
would hope is general enough to not be considered "domain specific".

Jim

On Mon, Feb 13, 2012 at 1:15 AM, Luc Moreau <L.Moreau@ecs.soton.ac.uk>wrote:

>  Hi Satya,
>
>  Your message is clear: you refer to the biomedical domain.  To me, this
> is domain specific.
>
>  Whereas,  "There are three types of agents in the model since they are
> common across most anticipated domain of use".  Furthermore, we say It is
> not an exhaustive list.
>
>  I would suggest that the best practice example should create a new class
> of agent that addresses a domain specific need.
>
>  This would be much more compelling, it would show we invite communities
> to define such subclasses, and it would show how to do it.
>
>  Do you want to help craft such an example?
>
>
> Professor Luc Moreau
> Electronics and Computer Science
> University of Southampton
> Southampton SO17 1BJ
> United Kingdom
>
> On 12 Feb 2012, at 23:36, "Satya Sahoo" <satya.sahoo@case.edu> wrote:
>
>   Hi Luc,
>
>>
>>  Of course we can talk about routers.
>>
> Exactly - there are many provenance-related scenarios in variety of
> application domains. Adding software agent to DM core will make it harder
> for users in say clinical research (majority use paper-based record
> keeping), bench research developing new vaccine targets (not using
> in-silico approaches) etc to adopt the model.
>
> Alternatively, is there a reason not to include both software and hardware
> agents? Is there any downside to include hardware agent, which is not there
> for software agent?
>
>
>>  But have had a use case, discussed by this wg and including routers?
>>
>>   Not sure what you mean - the wg is not discussing any "official" use
> case? We are using anecdotal scenarios to explain PROV constructs and not
> to drive creation of new constructs.
>
>  There are many biomedical use cases from XG and W3C HCLS group (e.g.
> mass spectrometer "hardware" and virus "biological" agents)?
>
>  A suggestion is to have two subtypes of agent (loosely from the
> provenance vocabulary approach)- biological and non-biological agents
> (hardware, software agents, organizations etc.).
>
>  What do you think?
>
>  Best,
> Satya
>
>
>>  Professor Luc Moreau
>> Electronics and Computer Science
>> University of Southampton
>> Southampton SO17 1BJ
>> United Kingdom
>>
>> On 12 Feb 2012, at 22:53, "Satya Sahoo" <satya.sahoo@case.edu> wrote:
>>
>>   Hi Luc,
>>
>>>  Nonhuman agent would imply other non software agents too.  It does not
>>> capture the intent.
>>>
>>>   Is the intent to model only software agents?
>>
>>>
>>> Software is particular relevant for the web. I don't see the problem
>>> with it. What use case do you want to support Satya?
>>>
>>>   From my original mail on Dec 07, 2011:
>>
>>  >Comment: Why should the WG model only these three types of agents
>> explicitly. What about >biological agents (e.g E.coli responsible for mass
>> food poisoning), "hardware" agents (e.g. >reconnaissance drones, industrial
>> robots in car assembly line)? The WG should either enumerate all >possible
>> agent sub-types (an impractical approach) or just model Agent only without
>> any sub-types. >The WG does not explicitly model all possible sub-types of
>> Activity - why should a different approach >be adopted for Agent?
>>
>>  "hardware" is equally relevant "for the web" (e.g. "router").
>>
>>  Best,
>> Satya
>>
>>
>>>  I had the feeling that we had reached agreement two months ago on this
>>> matter, and I don't see any new evidence to reopen the debate,
>>>
>>>
>>>  Ultimately we have to be pragmatic and move on.
>>>
>>
>>> Professor Luc Moreau
>>> Electronics and Computer Science
>>>  University of Southampton
>>> Southampton SO17 1BJ
>>> United Kingdom
>>>
>>> On 12 Feb 2012, at 20:23, "Satya Sahoo" <satya.sahoo@case.edu> wrote:
>>>
>>>   Hi all,
>>> I agree with Olaf's suggestion - its effectively captures our intent.
>>>
>>>  Thanks.
>>>
>>>  Best,
>>> Satya
>>>
>>> On Sun, Feb 12, 2012 at 2:20 PM, Paul Groth <p.t.groth@vu.nl> wrote:
>>>
>>>> Hi Olaf,
>>>>
>>>> That seems reasonable to me. I wonder what the group thinks.
>>>>
>>>> cheers,
>>>> Paul
>>>>
>>>> Olaf Hartig wrote:
>>>>
>>>>>
>>>>> Paul Groth<p.t.groth@vu.nl>  wrote:
>>>>>
>>>>>
>>>>>  Hi Satya,
>>>>>>
>>>>>> What's a good name for the class of both hardware + software
>>>>>> agent?
>>>>>>
>>>>>
>>>>>  In the Provenance Vocabulary we use the term NonHumanActor; so, maybe
>>>>> "non-human agent" for PROV?
>>>>>
>>>>> Cheers, Olaf
>>>>>
>>>>>   The key issue is that we need to distinguish between People and
>>>>>> Software so I this should be kept in the model.
>>>>>>
>>>>>> Thanks, Paul
>>>>>>
>>>>>>
>>>>>>
>>>>>> Satya Sahoo wrote:
>>>>>>
>>>>>>> Hi Luc, My suggestion is to: a) Either remove software agent or
>>>>>>> include hardware agent (since both occur together). b) State the
>>>>>>> agent subtypes as only examples and not include them as part of
>>>>>>> "core" DM.
>>>>>>>
>>>>>>> Except the above two points, I am fine with closing of this
>>>>>>> issue.
>>>>>>>
>>>>>>> Thanks.
>>>>>>>
>>>>>>> Best, Satya
>>>>>>>
>>>>>>>
>>>>>>> On Mon, Jan 16, 2012 at 5:40 AM, Luc
>>>>>>> Moreau<L.Moreau@ecs.soton.ac.**uk <L.Moreau@ecs.soton.ac.uk>
>>>>>>> <mailto:L.Moreau@ecs.soton.ac.**uk <L.Moreau@ecs.soton.ac.uk>>>
>>>>>>>  wrote:
>>>>>>>
>>>>>>> Hi Satya, Paul, Graham,
>>>>>>>
>>>>>>> I am proposing not to take any action on this issue, except
>>>>>>> indicate, as Graham suggested, that these 3 agent types "are
>>>>>>> common across most anticipated
>>>>>>>
>>>>>> domains
>>>>>>
>>>>>>> of use".
>>>>>>>
>>>>>>> I am closing this action, pending review. Regards, Luc
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On 12/07/2011 01:58 AM, Provenance Working Group Issue Tracker
>>>>>>>
>>>>>> wrote:
>>>>>>
>>>>>>> PROV-ISSUE-188: Section 5.2.3 (PROV-DM as on Nov 28)
>>>>>>>
>>>>>> [prov-dm]
>>>>>>
>>>>>>> http://www.w3.org/2011/prov/__**track/issues/188<http://www.w3.org/2011/prov/__track/issues/188>
>>>>>>> <http://www.w3.org/2011/prov/**track/issues/188<http://www.w3.org/2011/prov/track/issues/188>
>>>>>>> >
>>>>>>>
>>>>>>> Raised by: Satya Sahoo On product: prov-dm
>>>>>>>
>>>>>>> Hi, The following are my comments for Section 5.2.3 of the
>>>>>>>
>>>>>> PROV-DM
>>>>>>
>>>>>>> as on Nov 28:
>>>>>>>
>>>>>>> Section 5.2.3: 1. "From an inter-operability perspective, it is
>>>>>>> useful to define some basic categories of agents since it will
>>>>>>> improve
>>>>>>>
>>>>>> the
>>>>>>
>>>>>>> use of provenance records by applications. There should be
>>>>>>>
>>>>>> very
>>>>>>
>>>>>>> few of these basic categories to keep the model simple and
>>>>>>> accessible. There are three types of agents in the model: *
>>>>>>> Person: agents of type Person are people. (This type is
>>>>>>> equivalent to a "foaf:person" [FOAF]) * Organization: agents of
>>>>>>> type Organization are social institutions such as companies,
>>>>>>> societies etc. (This type is equivalent to a "foaf:organization"
>>>>>>> [FOAF]) * SoftwareAgent: a software agent is a piece of
>>>>>>> software." Comment: Why should the WG model only these three
>>>>>>> types of agents explicitly. What about biological agents (e.g
>>>>>>> E.coli responsible for mass food poisoning), "hardware" agents
>>>>>>> (e.g. reconnaissance drones, industrial robots in car assembly
>>>>>>>
>>>>>> line)?
>>>>>>
>>>>>>> The WG should either enumerate all possible agent sub-types
>>>>>>>
>>>>>> (an
>>>>>>
>>>>>>> impractical approach) or just model Agent only without any
>>>>>>> sub-types. The WG does not explicitly model all possible
>>>>>>> sub-types of Activity - why should a different approach be
>>>>>>> adopted for Agent?
>>>>>>>
>>>>>>> Thanks.
>>>>>>>
>>>>>>> Best, Satya
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> -- Professor Luc Moreau Electronics and Computer Science tel: +44
>>>>>>> 23 8059 4487 <tel:%2B44%2023%208059%204487> University of
>>>>>>> Southampton fax: +44 23 8059 2865 <tel:%2B44%2023%208059%202865>
>>>>>>> Southampton SO17 1BJ email: l.moreau@ecs.soton.ac.uk
>>>>>>> <mailto:l.moreau@ecs.soton.ac.**uk <l.moreau@ecs.soton.ac.uk>>
>>>>>>> United Kingdom
>>>>>>> http://www.ecs.soton.ac.uk/~__**lavm<http://www.ecs.soton.ac.uk/~__lavm>
>>>>>>> <http://www.ecs.soton.ac.uk/~**lavm<http://www.ecs.soton.ac.uk/~lavm>
>>>>>>> >
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>
>>
>


-- 
Jim McCusker
Programmer Analyst
Krauthammer Lab, Pathology Informatics
Yale School of Medicine
james.mccusker@yale.edu | (203) 785-6330
http://krauthammerlab.med.yale.edu

PhD Student
Tetherless World Constellation
Rensselaer Polytechnic Institute
mccusj@cs.rpi.edu
http://tw.rpi.edu