The Adaptive Generation and Coordination of Agents for Management by Exception
Management by exceptions is an effective methodology in business management. It suggests executives to focus their attention on potential problem areas (i.e. critical success factors) rather than being involved with every activity. A management information system that can support management by exceptions should be able to continuously monitor each critical success factor to check whether exceptions happen. In this project, we plan to design a model (called AGCAME) for building those executive information systems (EIS) that aim at satisfying the executives' information needs in management by exceptions. In AGCAME, a monitor corresponds to an autonomous agent with its own fundamental features including the priority, targets, constraints, and intervals of monitoring. Thus the executives may fully control the exceptions without wasting valuable time in repeatedly checking the huge amount of data. Since there will be many data items to be monitored, and each executive may have his own definitions and preferences of critical success factors, the agents should be adaptively generated and coordinated so that all useful exceptions may be efficiently detected. Therefore, based on our previous studies on adaptive information systems, we will design the adaptive way of generating and coordinating the agents to promote the practical feasibility of the agent-based EIS. The techniques from machine learning and natural language processing are integrated in the design. An EIS based on AGCAME may serve as a personal staff for the executive, since the behaviors of the EIS may be totally specified by and automatically adapted to the executive. The difficulty in system delivery for most management information systems is reduced, since the underlying transaction processing system is not changed at all. An EIS for managerial accounting and inventory control will be developed based on AGCAME. We will explore the performance of AGCAME both empirically and theoretically.
Keywords: Agent, Management by Exception, Executive Information System, Adaptive Generation of Agents, Adaptive Coordination of Agents, Machine Learning, Natural Language Processing
Back to Research Project