jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rolf Kulemann <r...@apache.org>
Subject Re: [PROPOSAL] Public NodeType Library (pntl)
Date Sun, 31 Oct 2004 11:48:58 GMT
On Sun, 2004-10-31 at 12:07, David Nuescheler wrote:
> hi rolf,
> > Interesting discussion here. I have to admit I'm not a semantic web
> > profi, but we/I in Lenya land have the idea to describe things like
> > workflow (and maybe other parts) with, eh, ontologies. We then would use
> > those _conceptual_ metadata to describe workflow of documents.
> > This is just a draft idea maybe only living in my head, and I'm not sure
> > if we should attach the wf meta data direct to document nodes or keep
> > them separate. We aren't that far in that discussion, yet.
> i think the workflow would be a very interesting topic to agree upon.
> since it has a very limited scope. i would also not call it meta information
> it is just application data in my mind and the application happens to be a
> "workflow" application.

Well, ok, I meant I have in mind to model the workflow "info data" as
meta data attached to special node types; lets say content or document
nodes. Maybe I have not enough knowledge about JCR, yet, and I'm wrong

My idea is to describe the concept of workflow using OWL/RDF. 

Lets say we have (OWL) classes: WorkflowEvent, WorkflowState and
WorkflowTransition. Instances of those are attached as meta information
to a document node, however. As I said, only draft ideas, since I have
to do other work ATM.

Using an OWL/RDF based approach would also make it possible to integrate
workflow info into other applications equivalence classes (right term?).
This bridges the gap between the _lowlevelness_ of JCR and specific
applications like CMSs. 

> currently we (day software) model our workflow in a fashion that
> is directly attached to the "document" (well, content really... ).

It has some advantages like if you move a doc, all closely related info
is also moved (workflow info). Otherwise you would have to maintain
links or such.

> i think it should not be too difficult to model a nodetype that could
> be directly attached as a mixin to a document or just refer to it.

Eh, ok, I have to read the JCR spec a bit more detailed. I'm really a
rookie and sometimes naiv.

> i would be very interested in having a discussion thread to 
> see if we can reach any consensus on a minimal "workflow" 
> nodetype that we could use as an example workflow model 
> in our applications... anybody interested?


Rolf Kulemann

View raw message