Intelligent Agents in Software Reuse Repositories

This article explores the role for intelligent agents in “reuse supported problem soving” in general, and in software reuse repositories in particular. As reuse repositories grow in size and extent, the “traditional” browse and surf mode becomes less effective, and simple search agents alone are not the answer. This article postulates a human information processing model of reuse supported problem solving, and then suggests a suite of cooperating agents that anticipate and better support the needs of repository users. Empirical results are presented from three separate repositories that compared the traditional vs. suite-of-agents architectures. These results show the intelligent agent architecture offers many benefits if agent parameters are properly tuned for degree of intervention and control, purpose and intentionality, emotional impact, and knowledge of reuse processes/domains. Readers will gain insight into (1) what agents should be included in the architecture and (2) effective settings of these agents’ parameters as a function of user skill and agent purpose.