Knowledge Representation Using a Modified Earley's Algorithm

Attribute grammars (AGs) have been proven to be valuable tools in knowledge engineering applications. In this paper, we formalize knowledge representation problems in their AG equivalent form and we extend the Earley’s parsing algorithm in order to evaluate simultaneously attributes based on semantic rules related to logic programming. Although Earley’s algorithm can not be extended to handle attribute evaluation computations for all possible AGs, we show that the form of AGs created for equivalent logic programs and the related attribute evaluation rules are such that allow their use for knowledge representation. Hence, a fast one-pass left to right AG evaluator is presented that can effectively be used for logic programs. We also suggest a possible software/hardware implementation for the proposed approach based on existing hardware parsers for Earley’s algorithm, which work in coordination with a conventional RISC microprocessor and can assist in the creation of small-scale applications on intelligent embedded systems with optimized performance.