Minutes are OK.

Barry Tao introduced himself to the group. He has a background in ontological matters, and has been with Southampton recently.

After straightening out a login problem, everyone seems good to go.

Discussion of OWL Clients

Several process discussions, about the clients that we are using. Primarily Protege (with OWL) will be our tool, because it is free and capable. (The latest version may have OWL support included.)

TopBraid has some nice features and will be used by some, but is not required.

Users will also need a subversion client to check out the ontology (and check in any changes we made).

The updated OWL file is at http://mmi.svn.sourceforge.net/svnroot/mmi/mmisw/device.owl. If you check this out with an SVN client, you will see the comments Luis added about his changes (copied here this one time for convenience):

• renamed class ObservableProperty for Property
• make processInput and processOutput subclass of Property
• renamed observable for feature
• renamed hasObservableProperty for hasProperty
• changed all labels to be lowerCamelCase
• removed hasObservable
• renamed class device for system

Some discussion of whether processInput/processOutput are types of properties.

Also discussed the term dimension -- why is it here, and what does it mean? Should property have a dimension? Not sure what functional value it adds. We'll delete that for now.

What is relationship of platform to the system? We are defining a system as a platform, sensor, or device; it has a platform. Here the system is a very generic device, perhaps too broad. Watch out for the terms 'system' and 'property', which may be reserved words in Protege. It may be that subclasses of systems should be created, and make this system an abstract class. Three examples: observing system, platform, sensor -- but these often have overlapping meaning and function.

What is relationship of deployment to system and process? System and process are similar; a system is a (physical?) extension of a process. If system is concrete thing, process has a time constraint, but system has multiple deployments over time. How do you explain partOf relationship between system and deployment, by using that it gives the impression of physical composition, as opposed to an associative relationship.

We are saying a deployment is a special type of process, but we don't have much information about that distinction.

need better examples, e.g., for process, in oceanography scenarios.

• Bruce Andrews
• Luis Bermudez
• John Graybeal
• Barry Tao
• Jesper Zedlitz