Re: possible SML use cases

For the first case, assume the following scenario:
1. The consumer makes a request to a data access service.  The  
consumer supplies the variables/properties/... for which values are  
needed and the vocabularies/semantic models ... that define them.
2. The data access service looks at the request and calls a service  
to look for equivalent requests in other vocabularies and sends the  
data source and vocabulary information back to the data access  
service.  The information could include pointers to processing  
instructions, say units conversions, that may be needed to generate  
appropriate values from the equivalent requests.
3. The data access service determines which sources to use and  
retrieves raw values from those data sources.
4. The data access service calls other services as needed to invoke  
processing instructions from step 2.
5. Other processing may be done to assemble a consistent response  
payload.
6. The response is sent back to the consumer.

In steps 2-5, the data access service (the one the consumer directly  
invoked) looks for appropriate sources and identifies appropriate  
processing to be performed.  But in step 2 there may be different  
sources available (or unavailable) or different options for  
equivalent requests when the same consumer request is made next  
week.  The processing in steps 4 and 5 may have changed.  While the  
response may still be valid, a change in the sources or component  
services may result in a response different from if the request was  
satisfied in an identical fashion each time.

Now sometimes having the new sources or services is exactly what is  
needed.  But if the response changes, the consumer needs to be able  
to evaluate why it changed.

Hope this clarifies my thoughts.

Ken


On Sep 17, 2007, at 1:36 AM, Pratul Dublish wrote:

> Hi Ken
>
> Thanks for your interest in SML and forwarding these use cases to  
> the WG.
>
>
>
> In the first scenario, it is not clear to me why you can’t  
> guarantee repeatability across different requests. I can understand  
> the service returning different data if the data sources have  
> changed, but this should be fine.  Please help us understand the  
> reasons that prevent repeatability in this scenario. Right now, I  
> am unable to determine if SML will be of any help here.
>
>
>
> In the second scenario, you should be able to capture the execution  
> context as an SML model  - defining  schema, inter-document  
> references, instance documents, and Schematron constraints to  
> capture the execution  context.
>
>
>
> SML is basically XML Schema 1.0 augmented with inter-document  
> references that capture relationships between documents, some built- 
> in constraints on inter-document references, and Schematron  
> constraints.  If you can use XML Schema 1.0 and XML 1.0 to capture  
> some aspects of these two scenarios, you should be able to use SML  
> to capture additional aspects of these scenarios.
>
>
>
> Thanks!
> Pratul
>
>
>
>
>
> From: public-sml-request@w3.org [mailto:public-sml-request@w3.org]  
> On Behalf Of Ken Laskey
> Sent: Saturday, September 15, 2007 6:01 AM
> To: public-sml@w3.org
> Subject: possible SML use cases
>
>
>
> I've exchanged email with Jim Lynn and he thought that these looked  
> like interesting use cases but suggested I post them to this list  
> to get a better feel for their applicability.
>
>
>
> So here they are.  Thoughts are welcome.
>
>
>
> Ken
>
>
>
> Begin forwarded message:
>
>
>
>
> From: Ken Laskey <klaskey@mitre.org>
>
> Date: September 15, 2007 8:45:27 AM EDT
>
> To: "Lynn, James (HP Software)" <james.lynn@hp.com>
>
> Subject: Re: Meeting minutes: 2007-09-14
>
>
>
> Jim,
>
>
>
> I don't know a lot about SML and have not had a chance to read the  
> spec since a pre-Submission draft but I'm interested in what its  
> use might be in several scenarios:
>
>
>
> 1. I send a request to a data access service asking for values for  
> certain properties, where I specify the vocabulary in which the  
> property semantics are defined.  The data access service finds  
> appropriate data sources and sends me back a response.  But a bare  
> response with the values is of limited use if I don't know where  
> those values came from or how information was processed to generate  
> those values.  it is especially a problem if I make the request  
> again next week and can't guarantee repeatability.  Could SML be  
> used as a logging mechanism to tell me unambiguously how the values  
> were generated?  Could I later submit that log and have the process  
> re-executed?
>
>
>
> 2. In the OASIS SOA Reference Model, the following is defined and  
> discussed
>
>
>
> The execution context of a service interaction is the set of  
> infrastructure elements, process entities, policy assertions and  
> agreements that are identified as part of an instantiated service  
> interaction, and thus forms a path between those with needs and  
> those with capabilities.
>
>
>
> Is SML a reasonable means by which to capture the execution context?
>
>
>
> Thanks for helping to pull some pieces together.
>
>
>
> Ken
>
>
>
> On Sep 15, 2007, at 8:30 AM, Lynn, James (HP Software) wrote:
>
>
>
>
> Ivan et al,
>
>
>
> I am an editor on the SML WG. Just as an fyi for your group, this  
> is SML 1.1, we are scheduled to release Second Public Working Draft  
> next week.
>
>
>
> Regards,
>
>
>
> Jim
>
>
>
> From: w3c-semweb-cg-request@w3.org [mailto:w3c-semweb-cg- 
> request@w3.org] On Behalf Of Ken Laskey
> Sent: Friday, September 14, 2007 5:10 PM
> To: Ivan Herman
> Cc: W3C SW Coordination Group; Kathryn Blackmond Laskey
> Subject: Re: Meeting minutes: 2007-09-14
>
> FYI, re Service Modeling Language (SML), see http://www.w3.org/XML/ 
> SML/
>
>
>
> Ken
>
>
>
> On Sep 14, 2007, at 11:16 AM, Ivan Herman wrote:
>
>
>
>
>
>
> ---------------------------------------------------------------------- 
> -------
>
> Ken Laskey
>
> MITRE Corporation, M/S H305      phone: 703-983-7934
>
> 7151 Colshire Drive                         fax:       703-983-1379
>
> McLean VA 22102-7508
>
>
>
>
>
>
>
>
>
>
>


------------------------------------------------------------------------ 
------------------
Ken Laskey
MITRE Corporation, M/S H305     phone:  703-983-7934
7515 Colshire Drive                        fax:        703-983-1379
McLean VA 22102-7508