Importance

Loading Importance Indicators...

Abstract

A computer implemented compilation method or compiler and translator that automatically converts an interpretive rule-based expert system into compact, compiled, parallel Ada code. The present compiler customizes the compiled code for each desired knowledge base to produce just the minimum amount of Ada code needed to provide equivalent functionality to one pass of the inference engine. After parsing, a data dependency analysis is performed on a rule base to identify segments that can be executed in parallel. These segments are then compiled by a code generator to make use of Ada's multi-tasking capabilities. The present invention accelerates the execution speed of the compiled code, reduces memory requirements for the compiled code, and enables expert systems to be efficiently embedded into real-time procedural systems, especially those using Ada software and parallel processing hardware. The present invention also includes code slicing techniques that are applied to rule bases that extends the translation capability of the present compilation method to encompass backward chaining in rule-based expert systems. In contrast to conventional approaches, the present approach provides a customized translation for each individual expert system, yielding minimal procedural code that is highly efficient in terms of both execution speed and memory requirements. This makes it more feasible to embed the expert system into a real-time application where high execution speed and low memory usage are critical.

Loading the Abstract Image...

First Claim