earticle

영어

In this paper we assume a group of collaborating agents where each agent is defined as an Information System. The system is coupled with a Query Answering System and a knowledge base which is initially empty. Information Systems can be incomplete. A user issues a query to the local database S in search for objects in information system that match a desired description. In case when some components of the description are missing in S, the query cannot be answered. The client then has to contact other agents to get some definitions of missing parts of the query. Our goal is to find optimal agent for client, where by optimal we mean an agent of maximal precision and recall. From all the agents we have, we choose the agent with the minimal value of distance, which corresponds to the closest agent to client. The definition of distance d discovered at the closest agent to the client and stored in KB of the client will guarantee that Query Answering System connected with the client has maximal precision and recall in group of agents having attribute d.