Towards a Framework for Symposium Planners
Rule Responder as an infrastructure for rule-based collaboration between distributed members of a virtual organization has seen several instantiations in the form of a Symposium Planner System. Although each instantiation employs three classes of agents Organization Agents(OA), Personal Agents(PA) and External Agents(EA) where the OA uses an owl ontology as a responsibility assignment matrix(RAM) to find a PA that can handle a request, with an enterprise service bus(ESB) to handle interactions between distributed agents, rule engines like PROVA and OOJDREW to for rule interchange utilizing Mule as a middleware between agent platforms yet the implementations have differed over the years in terms of agent technology.
New_Update.png


While we see 2007 to 2009 implementations of the framework employ the same infrastructure. The 2010 implementation of the Symposium planner used Emerald framework which is built on top of the JADE multi-agent system. The exercise was an inquiry into discovering alternatives to existing frameworks in an attempt to deploy trusted, third party reasoning services
instead.
Emerald.JPG

The Symposium Planner 2011 uses latest Mule ESB 3.0 coupled with support from Prova 3.1.3. Besides consulting knowledge from its repository the architecture also uses Semantic Web Dog Corpus a resource related to people, conferences and workshops related to Semantic Web research. The Symposium Planner has 3 Organizational Agents. A super OA delivers and filters queries to other sub OAs. Considering we have events of Rule ML 2011, this gave rise to the need for another OA to decide regarding sub OAs which were to be sent the filtered queries.
Keeping in view various implementations of the Symposium Planner, it seems a generalized implementation of Symposium Planner for various events is needed. The future implementation in question needs increased automation of processes as well as better support from human operators to add flexibility(the system already contacts human operator in case the query is not solvable by the agent itself). The need for present configuration involving 3 OAs(Organizational Agent) was to bring clarity in system operations.

OA_2011.png

However, with increasing usage of the planning system, such a scheme can become redundant. It is needed that we replace the 3 OAs with one OA which was part of the original implementation of the Symposium Planner. It can be achieved through unification of the business rules of the organizational agents which were decomposed for clarity at the cost of redundancy.
We can also add more human support to the system later in order to enhance flexibility. However, the initial improvements should be in the form of better support through systems. We can achieve more automation in terms of responses from agents as they replace actual human users and try to respond to increasingly complex queries. This might also lead to the need of communication between personal agents in order to help each other in answering queries posed by external agents. Personal agents use FOAF like profiles to advertise their capability of solving any query. This capability is used by the Organizational agents to decide the best Personal Agents to contact for a query. However, Personal agents cannot communicate with other Personal Agents to issue a solution. There is also the other way of looking at this dependency on the organizational agent, which can be made smarter by decomposing the query into various parts and contacting relevant Personal agents to solve those parts. The organizational agent can then assemble the various parts into a complete solution to a complex query.


Query_dec.jpg

In terms of future implementation of the Symposium Planner, we are looking at enhancements in automation and levels of human involvement to come up with a general framework that can take over the task of automating a symposium.