Agent Collaboration as a Resource Discovery Technique

Asking someone with experience in a particular eld is a natural way to learn about that eld. For example, students often ask teachers for suggestions for further reading. This paper describes an agent-based information system that explicitly supports such a learning method. Each user is represented in the system by an agent that contains a proole of its user's experiences and scripts that automate frequently-used searching patterns. Proole segments from one agent can be directly imported into another agent, thereby facilitating resource discovery and leveraging the costs associated with script creation.