Fast production systems for arbitrary architectures using LR parsing
Louis Paul Slothouber · 1987
W e present a method for automatically generating efficient production system interpreters for use on arbitrary architectures. Because of their need for speed and compactness, m a n y aeronautics and aerospace applications are restricted t o particular hardware and software environments (eg, embedded systems and AdaTM ). These applications cannot afford the symbolic processors and development environments usually necessary to execute large production systems efficiently. W e present an algorithm for translating any production system into an augmented context-free grammar. The concept of element classes is presented t o facilitate translation. A n existing program, the parser generator, creates parse tables which represent the original production system. These tables can then be used as data by a program that functions as a production system interpreter. This interpreter is at least as fast as that of other production system implementations (e.g., OPS5), while i ts storage requirements are much smaller. W e extend the notion of backward cha in ing . Those production systems t o which backward chaining is applicable notice a significant speed increase.