[Top] [All Lists]

Re: [ontolog-forum] Technical Discussion Session on "Semantic Web Servi

To: "Amit Sheth @ LSDIS" <amit@xxxxxxxxxx>
Cc: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
Cc: "verm >> KUNAL VERMA" <verma@xxxxxxxxxx>
Cc: David Martin <martin@xxxxxxxxxx>
From: Nicolas F Rouquette <nicolas.rouquette@xxxxxxxxxxxx>
Date: Fri, 28 Oct 2005 10:24:33 -0700
Message-id: <43625ED1.9070907@xxxxxxxxxxxx>
At JPL, we are also familiar with the 'data' problem; however, since we 
are engineers,
we see things a bit differently and distinguish two kinds of 'data':    (01)

- 'data' as a 'product' (we call that a 'data product')    (02)

In the sense of Barry Smith's recent presentation on ontolog, an 
autographic document (i.e., the product itself)
that has an underlying allographic document (i.e., the information 
expressed in that product)    (03)

- 'data' as the 'state' of something    (04)

Again, there's a conceptual distinction between an autographic 
description of state (which we call a 'state view')
vs. an allographic descriptio of state (which we call a 'state function' 
because state is fundamentally dependent on some
notion of 'time' whether it is physical time, mythical time or some 
other indexing criteria used as 'time').    (05)

There's more about this stuff in the paper I've attached.
My involvement in that work as been as the implementation
level where I spent a great deal of time & energy into couching
that work in an explicit state architecture model. (no shameless plug to 
give you though).    (06)

-- Nicolas.    (07)

Amit Sheth @ LSDIS wrote:    (08)

> Nicolas:
> Very interesting points, which I agree with to a good extent.
> Some  minor caveat/clarification:
> 1. UGA-LSDIS's work is not only limited to WSDL-S.
> METEOR-S has explores many issues since 2000 or 2001
> (design, annotation, publication, discovery, composition, execution 
> with monitoring and measurements;
> semantics covering data, behavior/function, non-functional/Qos, 
> execution, as
> we have discussed since 2003; papers available from the LSDIS library)
> beyond the higly targeted approach
> of WSDL-S.  WSDL-S target is motivated by the fact that from a 
> standardization point of view,
> one needs to take a small, relatively easy (technical as well as 
> training/engineer's comfort level- wise)
> steps with common denominator that is likely to be acceptable to a 
> community (here defined by
> W3C members; a practical minset expressed in one of the recent posting 
> by W3C team member
> incidated interest in specification that extends WSDL2.0; there are 
> many issues covered by
> some of SWS initiatives/submissons that deal that go well beyond the 
> scope of WSDL;
> this is not to say they are not important or that they cannot be 
> addressed in a longer term in
> an extensible manner).
> 2. Much of the formal representation/modelingwork is about "control" 
> aspect of the process/composition/workflow;
> the bigger challenge is the "data" aspect.  We have also done some of 
> the early (early 90s) formal
> work related tothe control aspect of workflows/processes, some ideas 
> from it are still quite relevant in my biased opinion 
> But when I built a commercial workflow management system METEOR AppS 
> from Infocosm
> (based on licensing METEOR workflow technology from LSDIS-UGA) and 
> applied to real-world
> applications (e.g., neonatal clinical pathway support system), most of 
> the control
> specifications (patterns) issues were easy (none challenging).  On the 
> other hand, it took
> an order of magnitude more effort and time to deal with data 
> heterogeneity/mapping.
> And in those days we were still doing things that were 
> intra-enterprise; now with high degree of
> autonomy in service development and more loosely coupled nature,
> we face more heterogeneity.  There is a lot of early (e.g., 
> http://www.dagstuhl.de/files/Proceedings/04/04391/04391.SWM2.Other.htm)
> and recent work on data issue which
> folks on services/process sides have not taken into consideration 
> (albeit sucm such as
> WSMO and METEOR-S are trying).  Yesterday a student just reviewed (as 
> part of our
> effort to develop use cases and showcase scenarios for WSDL-S):
> -- an existing web service and its semantic annotation using an 
> existing ontology
>    (did not seem to me a bad example at all; we will package this with 
> RADIANT tool
>      which enables WSDL-S annotation)
> -- semantically enriching the existing ontology and doing the above, 
> resulting on
>   more useful mappings
> In my view, difficult semantic issues are in matching (with 
> disambiguation) and mapping
> and WSDL-S provides a "platform" to *start* addressing these. To my 
> knowledge,
> as part of their data mediation, WSMO is looking at these
> challenges too, but I feel all efforts/proposals need to give this 
> issue more
> importance.
>> Even on the practical level, there are significant challenges to
>> layer a computational model that is adequate for implementing SWS on 
>> top of >leading-edge engineering platforms like grid-scale
>> computing or the next-generation internet for space exploration.
> Agreed. EU funded ASG 
> (http://asg-platform.org/cgi-bin/twiki/view/Public) is one of the 
> projects looking at this,
> and as a shameless plug, we just got funded SemGrid-- a project that 
> supports METEOR-S - ASG collaboration
> (http://lsdis.cs.uga.edu/projects/semgrid/). But lot more is to come 
> here.
> Amit
> Nicolas F Rouquette wrote:
>>Amit, as far I'm concerned, foreign travel is an expense that at this
>>time would be very difficult for me to get approval for.
>>In a strange way, the scarcity of funding for SWS-related work makes it
>>hard for some people in the US to travel to Europe
>>and elsewhere where SWS is in heat.
>>In answer to Michael's question, I am particularly interested to hear
>>about 'shameless plugs' about R&D technology that could be
>>applicable in the context of JPL and to a larger extent NASA. However,
>>my criteria of 'applicability' is somewhat different than
>>the conventional notion of practical = tool-supported technology that is
>>simple enough to support agile methodologies.
>>My notion of pragmatism is an approach that addresses two ends of the
>>SWS spectrum:
>>- the practically agile side of tool-supported SWS methodologies
>>- the extremely formal side of theory-grounded SWS methodologies
>>If I can exagerate a bit, Amit's approach with Meteor-S is an example of
>>the first case
>>whose strength is in its developper-friendly approach with reasonable
>>tool support.
>>Michael Grunninger's approach is an example of the second case whose
>>strength is
>>in the precise and rigorous manner in which one can apply
>>theorem-proving techniques
>>to analyze, validate and certify one's approach. WSMO is somewhere in
>>So, one logical next step is to strategize about how we can make
>>stronger connections across
>>SWS methodologies so that we could, e.g., formally reason about a
>>practical SWS application
>>using an precisely defined model of the application processes according
>>to a formal ontology
>>of services in FLOWS + specific extensions according to the application
>>(e.g., are the activities deterministic or non-deterministic, are the
>>preconditions for activities
>>stated in terms of state, time, state+time, the occurence of other
>>activies, something else?,
>>do states have the Markov property (I.e., no history) or not?, ...)
>>The practical view of SWS is in some ways narrow focused to one thing:
>>using SWS as a way
>>of building an application. It's as if SWS were a new 'implementation'
>>language like Java or C.
>>This is a very narrow minded view of SWS in my opinion for 2 reasons.
>>1) If we have a precisely characterized formal ontology for a SWS
>>(i.e., something like FLOWS + all of the appropriate PSL trimmings)
>>then we could match the SWS application to other process formalisms like
>>PiCalculus or
>>other models of automata. These formalisms offer alternative methods for
>>analyzing an SWS application,
>>e.g., for purposes of model checking, reachability, commandability,
>>diagnosability, etc...
>>2) Even on the practical level, there are significant challenges to
>>layer a computational model
>>that is adequate for implementing SWS on top of leading-edge engineering
>>platforms like grid-scale
>>computing or the next-generation internet for space exploration.
>>-- Nicolas.
>>Amit Sheth @ LSDIS wrote:
>>>Would you or someone from the group be interested in offering a
>>>panel/follow up at ICSOC-DWP
>>>(http://lsdis.cs.uga.edu/projects/meteor-s/dwp2005/), or another
>>>workshop (mainly Mediate'05).
>>>If so, I can look to see if there is a slot.
>>>Michael Maximilien wrote:
>>>>Let me start by seconding Peter and thanking all the presenters and
>>>>to Nicolas for moderating.
>>>>I was wondering how we could follow-up on the discussions we have
>>>>started during the panel?  In particular, does it make sense to have
>>>>some workshop or some other form of discussions?  One approach is to
>>>>meet during an existing conference, e.g., the up coming ICSOC 2005
>>>>conference in Amsterdam in December of this year?
>>>>I am particularly interested in business applications of SWS and the
>>>>scenarios where the technologies are applicable.  In some sense, I
>>>>see the basic vision of SWS as not only a path to help automate some
>>>>aspects of SOA solutions but from a broader view to help move IT
>>>>services and processes (engagements) to the Web.
>>>>I am thinking of how IT service providers (e.g., IBM Global Services,
>>>>Accenture, and so on) currently execute service engagements or
>>>>particularly the kind of collaboration, knowledge, cocreation of
>>>>value, and so on, that occurs in a service engagement.  With the Web
>>>>as the service delivery platform, SOAs are allowing a dynamic
>>>>substrate for IT service providers and consumers.  However, as we all
>>>>know, one thing that is missing from current approaches, to enable
>>>>better automation, is a conceptualization of the semantics in the
>>>>domain and the services as well as tools, techniques, and approaches
>>>>to take advantage of that formalization.   And of course, that is
>>>>exactly what the SWS community has been researching and continues to
>>>>(This is also very much in line with a broader push by IBM research
>>>>to create a science of services.   See the SSME site =>
>>>>SHAMELESS PLUG: Peter and I and working to get Jim Spohrer to present
>>>>IBM's SSME research efforts to the ontolog-forum on December 8th)
>>>>However, an apparent comment is that the SWS vision seems like a
>>>>daunting task to achieve; and it begs for pragmatic and evolutionary
>>>>approaches...  hence some of the comments of making sure to align
>>>>closely with the Web services industry efforts and some of the
>>>>praises for the WSDL-S initiative.
>>>>Nonetheless, maybe the various threads and approaches are necessary
>>>>for progress and we should question the current assumptions
>>>>underlying the Web services standards.  The REST and Web 2.0 approach
>>>>to implementing Web services certainly seems to be doing just that
>>>>and getting good traction...
>>>>Where do we go from here?
>>>>Thoughts, suggestions, comments, questions, or feedback in general is
>>>>E. M. (Max) Maximilien, Ph.D.
>>>>IBM Almaden Research Center
>>>>San Jose, CA USA
>>>>Homepage: http://maximilien.org
>>>>"Peter P. Yim" <peter.yim@xxxxxxxx> wrote on 10/20/2005 06:16:50 PM:
>>>>>We had an most wonderful virtual session today with Nicolas
>>>>>Rouquette from NASA/JPL who moderated the Discussion on "Semantic
>>>>>Web Service Ontology Standard".
>>>>>Thanks to our distinguished panel, which included David Martin
>>>>>(OWL-S), John Domingue (WSMO), Amit Sheth (WSDL-S) and Michael
>>>>>Gruninger (SWSF / FLOWS), and the experts in the audience, we
>>>>>collectively delved into the various perspectives and approaches
>>>>>being taken by the state-of-the-art researchers on the subject
>>>>>matter. We tried to make some sense of the current situation, and
>>>>>even suggested where things should possibly be heading. Please
>>>>>refer, again, to details at the session wiki page at
>>>>>ConferenceCall_2005_10_20 where the presented materials and a lot
>>>>>of relevant resources have now been posted for everyone's reference.
>>>>>The digitized (mp3) audio recording of the entire discussion has
>>>>>also been posted. As usual, this will be archived in our
>>>>>knowledge repository for reference and download by anyone
>>>>>interested. Additionally, for the next 10 days, telephone
>>>>>playback will also available. See details at:
>>>>>Thanks & regards.  =ppy    (04)
>>>>>P.S. Our next major events is scheduled for Thursday, Nov. 3,
>>>>>2005,  Mr. Rex Brooks from HumanMarkup.org will be moderating a
>>>>>Take-II (as a follow-up to our 2005.08.25 session) of the Panel
>>>>>Discussion on "Healthcare Informatics Landscapes, Roadmaps, and
>>>>>Blueprints: Towards a Business Case Strategy for Large Scale
>>>>>Ontology Projects". Please mark your calendars ... and, watch out
>>>>>for our announcements on these two events which should be coming
>>>>>out any moment now.
>>>>>P.P.S. In view of the fact that the National Center for
>>>>>Ontological Research (NCOR) will be having their Inaugural Event
>>>>>next Thursday (at Buffalo, NY), where a good number of our
>>>>>community members will be present at, we will NOT be having our
>>>>>regular weekly conference call that day (repeat, no call on
>>>>>2005.10.27). So, talk to you again Thu 2005.11.03   =ppy
>>>>>Peter P. Yim wrote Fri, 14 Oct 2005 10:04:29 -0700:
>>>>>>We are pleased to announce our Technical Discussion session
>>>>>>for next Thursday, Oct. 20, 2005. The topic to be discussed will be:
>>>>>>"Semantic Web Service Ontology Standard". We are expecting a
>>>>>>distinguished panel and some of the top researchers and
>>>>practitioners at
>>>>>>the session. We will delve into the still divergent perspectives and
>>>>>>approaches being taken on the subject matter during this session.
>>>>>>Collectively, we will attempt to make some sense of the current
>>>>state of
>>>>>>affairs, and possibly suggest where things should be heading.
>>>>>>*Conference call-in details*:
>>>>>>Date: Thursday, Oct. 20, 2005
>>>>>>Start Time: 10:30am PDT / 1:30pm EDT / 17:30 UTC (World Time:
>>>>>>Session Duration: 2 Hours
>>>>>>Dial-in Number: +1-702-851-3330 (Las Vegas, Nevada, USA)
>>>>>>Participant Access Code: "686564#"
>>>>>>VNC shared-screen support available
>>>>>>* Topic: "*Semantic Web Service Ontology Standard*"
>>>>>>* Moderator: Nicolas Rouquette (NASA/JPL)
>>>>>>* Organizing Team:
>>>>>>     o  Nicolas Rouquette (NASA/JPL),
>>>>>>     o  John Domingue (KMI, UK), &
>>>>>>     o  E. Michael Maximilien (IBM)
>>>>>>* Panelists:
>>>>>>     o  David Martin (SRI) - OWL-S
>>>>>>     o  John Domingue (KMI, UK) - WSMO
>>>>>>     o  Amit Sheth (U of Georgia) - WSDL-S, and
>>>>>>     o  Michael Gruninger (U of Toronto) - SWSF / FLOWS
>>>>>>*Refer to details on the session wiki page at:*
>>>>>>This will be a virtual session over an augmented conference call.
>>>>>>The entire session will be recorded and made available as open
>>>>>>under the prevailing Ontolog IPR policy (see:
>>>>>>I look forward to having you at this open session. Please pass this
>>>>>>announcement along to those who might be interested to join us
>>>>too. All
>>>>>>are welcomed.
>>>>>>RSVP by by emailing me at <peter.yim@xxxxxxxx> offline.
>>>>>>Regards. =ppy
>>    (09)

Attachment: StateAnalysisMDSpaper.pdf
Description: Adobe PDF document

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (01)

<Prev in Thread] Current Thread [Next in Thread>