Description

Phosphorus is a knowledge-based matcher that accepts a user's description of a needed service as input and responds with a ranked list of agents that have the capability to provide that service. The Phosphorus matcher is both knowledge based and experience based. It accesses a base of ontological knowledge to assist the process of performing structural matches between pairs of requested and advertised capabilities, and it takes a knowledge-based approach to guide a post-match decision process that determines how retrieved capabilities can be selected, reformulated, and/or combined to best satisfy a user request. We assume that exact matches will be relatively rare; our matcher exploits domain ontologies to allow loose-coupling between agent capabilities and user requests, finding agents that have more specific capabilities than what was requested, or agents that can provide a related service when no exact matches are found. Loose coupling is enabled by using EXPECT's goal and capability language to describe agent capabilities and requests are then be matched through subsumption, goal reformulation, and partial match. The matcher is also experienced-based, using learning techniques to improve the utility of its matches over time.

Phosphorus is able to abstract the results of its learning, storing its experience in the form of axioms in a knowledge base that summarize its knowledge of matching requests with agents within a particular domain. Phosphorus can be used standalone or be used as a service within existing agent integration architectures.

More (Research)