Modeling a Distributed Knowledge Management for Cooperative Agents

Abstract

This paper shows modeling a distributed knowledge management in collaborative agents, introducing knowledge migration between agents. In multi-agent environment, each agent can be working at common goals with globally cooperative behaviors. In order to form cooperation, agents first migrate related knowledge, locally evaluate the others' requirements, then agents finally can form a plan to achieve goals, In order to construct integrating agent's behavior and cooperation among agents in knowledge level, we propose four goals; (I)cooperation of software agents in the communication between agents to achieve a common goal, (2)adaptability of a software agent when its autonomous software agent can control its correct behavior in the environment and can manage both its knowledge and other agents' migrated knowledge to execute its behavior in knowledge-level, (3)mobility of a real-world mobile agent when mobile computer and autonomous mobile robot equipped with a network can execute its behavior by knowledge migration between mobile computer and autonomous mobile robot, and (4)transparency of knowledge migration when the communication requires to construct transparent knowledge boundaries between real space and virtual space which a computer generates in its display. In this paper, we present two approaches for agent collaboration to resolve the above mentioned issues. As for the first approach, we introduce social agency model for constructing a prototype system for guide activities in a laboratory. As for the second approach, we present an autonomous agent's architecture in social agency aimed at communicating with other agents in knowledge-level. Based on these approaches, we will propose a distributed knowledge model between agents and within one agent.