Proposal of Profile and Event Sharing by Agent Communication

Finding the valuable data among huge amount of data is too laborious due to the human limitation on perception capability. Reversely, pushed data about a service must miss the point without understanding user's demand. So, our base idea is to collect profiles of users and providers, and to match the user's demand with the providers' service. In this paper, we propose a communication framework for agents to find friend-agents and to share events in place of the client. Concretely we propose two-stage protocols, that is, profile sharing and event sharing. Thereby, only significant events can be shared in real time since friend-agents already have logical linkage. We verify that the latency to convey the valuable event to the receiver can be reduced. Finally, we discuss the structure of the profile data, and shows that required load for word matching can be saved by the hierarchical structure.