Inference strategies for expert database systems

It is shown that for certain stable textual databases, specific inference strategies such as record clustering, semantic nesting, probabilistic ranking, and global constraint analysis can be used to enhance the performance of rule-based expert system front-ends for such databases. More specifically, we discuss the design strategies behind the inference engines of two expert system prototypes. The systems gfCrystal and gfGuru were developed as front-ends for a graduate fellowships database maintained at Concordia university. The system gfCrystal was developed using the expert system shell Crystal and uses a mixture of forward and backward chaining inferences. It was ported to the richer integrated programming environment Guru and called gfGuru. The latter system constitutes the core of a comprehensive student and employer information system currently being developed