[Swift-user] Virtual data schema / catalog?

[Yong Zhao]

Allen,

There is a in-depth paper that talks about provenance issues in VDS/Swift,
you can find it here:

http://people.cs.uchicago.edu/~yongzh/pub/IPAW_VDSProvenance.pdf

I also want to mention that I had implemented VDC (for the former VDS) on
top of eXist, a very good native XML database, which is far better than
Xindice. The XML based implementation could potentially support all the
queries we mentioned in the provenance paper.

For swift, the organization of the workflow is a bit different, as there
is not an explicit graph representation for a workflow, but the swift
runtime can generate a graph for a specific workflow instance (i.e. the
workflow is operating on a specific set of inputs).

As for human intervention, there is a BPEL4PEOPLE spec that talks about
integrating human interactions into workflows, but I don't know the
details.

Yong.

On Fri, 14 Sep 2007, Allen, M. David wrote:

> Hello,
>
> I'm just trying to find my way through the background on this software,
> and thought see if anyone could point me in the right direction.
>
> I first came to swift through reading the original Chimera paper from
> 2002.  The ability to distribute jobs across different machines is less
> interesting to me than the idea of tracking provenance for complex
> creations.  Is there a way that I can get this kind of structured
> provenance information out of swift, or a related tool?  Ideally, it
> would be structured similarly to the virtual data schema I saw in the
> 2002 Chimera paper.
>
> I have taken a look at the KML & XML files, and they seem to be more or
> less fit the bill.  (KML seems to have derivations -
> invocation-specific information that's more or less a direct
> translation of the swiftscript, and the XML file seems to have generic
> procedure metadata, the transformations)  Are there any tools available
> that can manipulate this information for other purposes (or insert it
> into a relational database with a proper schema?)
>
> If I'm overlooking some documentation, please provide a pointer. I'm
> more than happy to RTFM.  :)
>
> Thanks
> --
> M. David Allen
>
>