An adaptive agent for case description in diagnostic CBR systems

Case-based reasoning (CBR) systems can support diagnosis of complex industrial systems. The success of a diagnostic CBR system depends on its ability to retrieve previous cases that provide information to solve a new case. To this end, the new case must be adequately described. However, to describe a new case in an ill-structured diagnostic decision environment requires considerable domain knowledge and is dependent on the strategies used by a decision maker. In this paper, we develop a framework for the development of an adaptive agent that can assist a decision maker describe a new case to a diagnostic CBR system. The adaptive agent is dynamic and provides its recommendations based on the diagnostic strategy of a decision maker. An empirical evaluation of the proposed framework in the diagnostic of complex industrial machinery supports its effectiveness.