A computer hardware configuration expert system: Examination of its software reliability

Describes the implementation of an expert system called SCAT for a computer hardware configuration task. It assists system engineers in enumerating all the parts necessary to build a computer system and prepares the documents for succeeding tasks. The emphasis is placed on the software reliability metrics and the development management methodology employed in SCAT. To evaluate the software reliability of SCAT, a traditional metric was employed the number of faults detected at each stage of the test process was counted and compared with the standard value. By analyzing the fault data derived from two versions of SCAT, it was found that the code review for production rules was not as effective as it was for procedural languages. It was also found that there was a possibility that the number of faults embedded in production rules in the coding process was smaller than that in procedural languages