Re: [regrep-semantic] Semantic ebXML Registry Proposal: The TwoBindings Solution

[Farrukh Najmi <Farrukh.Najmi@Sun.COM>]

Zach thanks for writing our first high level proposal for SCM.

Chiusano Joseph wrote:

>Zachary Alexander wrote:
>  
>
>>Semantic Content Management Team:
>>
>>Here are some brief ideas I have for adding semantic capability called,
>>"The Two Bindings Solution."
>>
>>Problems:
>>• How to support Semantic capabilities and limit the number of changes
>>to the ebXML Registry Standard?
>>    
>>
>
>First I think we should ask the question of why the numer of changes
>should be limited - not to say that they shouldn't be of course, but to
>first step back and assess the current environment.
>  
>
+1

Also there is no telling what the changes would be in any approach until 
we dig deeper.

The first order bit question is what are the requirements we need to 
meet before we can
consider specific solutions.



>>• How to reduce the impact of supporting semantic capabilities by
>>current ebXML Registry vendors?
>>    
>>
>
>Along the same lines, I think we should first ask how many current
>vendors are there (I know that we know this), and then assess what the
>impact of supporting semantic capabilities might be to them, before we
>determine that the impact needs to be reduced.
>  
>
+1 again. There are very few registry vendors. Further, we could make 
the SCM
features initialy optional to reduce burden on vendors. In a subsequent 
release
we could make them mandatory.

>Joe
>  
>
>>Solutions:
>>• Create a Knowledge Interchange Format Binding.
>>        o Normative solution
>>• Create an Open Knowledge Base Connectivity Binding.
>>        o Non-Normative solution
>>• Make registry applications responsible for inference and query
>>formatting.
>>    
>>

This is a good example of why requirements need to be hammered out first.

I am assuming (needs validation) that the main point of SCM is for the 
registry
to be able to do inference based on knowledge bases (ontology + instance 
data).

The idea of leaving inference to applications seems to miss the mark by 
a lot.

For example it is not clear to me how would we be able to meet the 
Semantically Aware Query user case:

http://www.oasis-open.org/committees/download.php/5582/ontologySearch.html

without having the registry do inference.

I belive our goals are much higher than simply referencing Ontology 
classes with RegistryObjects.
I believe we need to be able to store ontologies, knowledge bases and be 
able to do inference based
on them.

Maybe its is time to finish our pass over use cases and scenarios and 
then move quickly into
establishing initial first pass requirements.

Can we start threads on suggested requirements based upon existing use 
cases and scenarios
as a next step?

-- 
Regards,
Farrukh