On Nov 27, 2007 9:42 AM, Andrew Hume <[EMAIL PROTECTED]> wrote: > not really. > > the idea is that to an amazingly great extent, you should be able to > automate b) completely. > and in a), the two models would mesh already (because there is one ubermodel > a la CIM). > >
Have you tried experimenting to see a minimally complete type system for the evolution and learning of a systems administration ontology? I worked on a very, very simple system that looked like this: Basic types: * Software (includes both the actual programs and configuration information) * Users * Machines (with subtypes according to the role: eg, regular host, network device, etc) with operations * Delete * Read * Infer * Verify * Edit * Create and top level policies that held * SLA * Constraint (e.g., RBAC). SLA can be considered a sub-type of a Constraint, but it's fairly central to how SAs think, that I made it a separate type -- I might want to reconsider that at some point. and from there worked on the minimal set of additional features (e.g., grouping, enumeration) required to accomplish the Use Cases that were discussed during a CM workshop at LISA. If you start with a minimal system (e.g., all components share the simple model above) that is extensible and composable, you may be able to do some of the distributed model of ontology negotiation & discovery that you're talking about. Type Theory (e.g., Behavioral Subtyping and Structural Subtyping) might be of interest to you there. Starting from a rich, shared ontology is very helpful, though. Steven _______________________________________________ lssconf-discuss mailing list lssconf-discuss@inf.ed.ac.uk http://lists.inf.ed.ac.uk/mailman/listinfo/lssconf-discuss