Re: RDF ontology for describing reusable software components

All,

Thank you very much to everyone who responded.  Forgive me for
addressing all points raised at once.  With luck, it will read faster
than an Oscar acceptance speech...

Charlie Abela wrote:
> This is an interesting idea, what worries me is the idea of 
> "enabling automated inference about how to assemble components to
> produce a solution to a stated goal". 
> Would you explain this statement, this has a lot of hidden issues?

Charlie, thank you for asking the question.  Certainly I have no desire
to worry anyone.  :) For more information about the goals of my project,
please visit http://oxygen.lcs.mit.edu/Software.html, and scroll down to
Approach/Specifications.  

As you can see, one of the goals is to support the downloading and
installation of communicating software modules to multiple local
devices.  In order to do this, it must be necessary for a system service
to:
1) Reason about whether the user's goals can be accomplished using the
components currently installed.
2) Identify any currently uninstalled modules that must be installed.
3) Assemble the resulting collection of modules to create a complete
solution to the stated goal.

This is in addition to the needs addressed by some Web Services
initiatives, which include the discovery and invocation of services that
remain remote.  Thus, we are not in the enviable position of having an
HTTP layer that separates us from the operational needs of the
component, including computational resources, software platform, and
required collaborating components.  However, we do not want to overlook
the work done on interfaces for Web Services already.

Charles McCathieNevile wrote:
> [snip]
> There are some folks at MIT intersted in describing Web services in RDF -
> [snip]
> Sandro Hawke or Eric Prud'hommeaux might be able to tell you
> more about this.

Charles, thank you for the names.  I hope to get in contact with them
soon.

> I also hope to have a student working with me this northern summer on a
> similar project to yours, for the purpose of describing security monitoring
> components.

Not sure when the "northern summer" starts, :) but please have your
student get in touch with me then, and we'll talk about possible
overlaps in concerns.

Pasqualino Assini wrote:
> we have been successfully using RDF to describe software components
> interfaces for more than two years in our Nesstar system.

Pasqualino, thank you for the link, and a very natural map for method
signatures.

Mark Baker wrote:
> I suppose my "RestRDF" thought experiment could be considered related,
> with Web resources being distributed components exposing a common
> application interface.

Mark, thank you for introducing me to REST.  I must say that it's a very
catchy idea.  Based on my further elaboration above, what do you think
are the places that REST can best inform our endeavors? 

Thanks again to all,

   David Saff

Received on Monday, 1 April 2002 19:18:50 UTC