ADAPTIVE AGENTS FOR EFFECTIVE INFORMATION MONITORING

Abstract

Information motoring is an essential basis for management and decision making in various domains. Once an information update is detected, suitable procedures may be triggered to identify potential problems and opportunities. In this paper, we define effective information monitoring (EIM) and explore how adaptive agents may cooperate with each other in order to achieve their collective goal of EIM. Since each information item may be updated by various entities (e.g. information servers) at any time, EIM calls for a multiagent system that may detect more information updates in a timely manner using a controlled amount of system resources (e.g. loading of related information servers and the Intranet). To achieve EIM, the agents should adapt themselves to the dynamically changing behaviors of the information items being monitored. They learn to issue requests of using system resources and concede to those agents that are more likely to detect updates at the time of negotiation. The framework is theoretically and empirically evaluated. Its potential applications to management by exceptions are identified and discussed.