NEURAL-NETWORK-BASED AGENTS FOR INTEGRATING INFORMATION FOR PRODUCTION SYSTEMS

Recent research on neural network-based agents for coordinating and fi ltering information is presented. Neural net agents have advantages ov errule-based agents because they can be easily adapted to the informat ion needs of the user through the learning capabilities of neural netw orks. Three application areas for these agents are discussed: electron ic bulletin boards, Group Decision Support System (GDSS) meetings, and electronic mail (e-mail). This paper reports accuracy results of the network for electronic bulletin boards and meeting messages. The agent s could accurately filter messages, under optimal conditions, in the r ange of 85% to 99%. Interface designs for incorporating these agents i n GDSS and email systems are discussed and illustrated. The advantages of neural network-based agents over rule-based agents are discussed a lso.