CMU

Dr. Katia Sycara

1. CMU will provide relevant ontologies, if appropriate, and also DAML-S descriptions of the profiles of the services that would be used in the Experiments. We view such services to inlcude weather services, information from battlefield "sensors" and even targets themselves, e.g. bridges, factories, etc (see a munitions to targets scenario below). These services will be registered with a matchmaker agent.

2. CMU will provide matchmaking algorithms to match DAML-S profile descriptions of service providers with DAML-S descriptions of service requests. Obviously, the matching can be partial (we define different categories of matching) and is based on matching concepts rather than keywords.

3. CMU will provide protocols expressed in the DAML-S grounding to allow requesters for services to interact with service providers.

Now, I was thinking that as part of the scenario we could have the following little snippet.

a. Targets e.g. bridges, (that of course are nominated  dynamically) describe themselves in DAML-S (i.e. providing a DAML-S profile) and register with a matchmaker.

b. Munitions that might be capable of attacking a particular target act as "requesters" for that type of (or instance of ) target through a DAML-S request to the matchmaker.

c. The match maker matches requests to registrations and gives the requester (and the human user) candidate targets that match the capabilities of the weapons.

d. One could go even further and have the weapons have a process model of the process by which they can be destroyed. The weapon that is assigned to a target, (after finding the target through the matching process), reads and understands the process model of the destruction of the target and "interacts" with it by taking the appropriate steps (modelled as exchange of messages) to destroy the target.

e. At the end of the interaction, the target  sends a message to signify "I have been destroyed" or "I have been partially destroyed", or "I have not been destroyed" depending on the assumptions of the overall scenario.

This scenario snippet can follow directly after a user's decision whether to destroy a target (based on the COA assessment and operational net assessment we have been discussing in the telecons