| Abstract: | The influence of the organization of a declarative knowledge base on the development and application of proceduralized knowledge was investigated in a complex troubleshooting domain. Domain explanations were generated in either a depth-first or breadth-first manner for different groups of subjects who were also given experience learning to troubleshoot in the domain. Although the two explanatory structures led to similar training performance, the two groups differed significantly in their overall level of performance in subsequent troubleshooting problems. Examination of objective measures of troubleshooting performance and think-aloud protocols indicated that breadth-first declarative knowledge representation fosters the use of mental models during problem-solving in training. It also facilitates proceduralization of that knowledge into fast and accurate methods for localizing faults. |
