An Efficient Context-Free Parsing Algorithm for Natural Languages

Abstract

This paper introduces an efficient context-free parsing algorithm and emphasizes its practical value in natural language processing. The algorithm can be viewed as an extend ed LR parsing algorithm which embodies the concept of a " graph -structured stack." Unlike the standard LA, the algorithm is capable of handling arbitrary non-cyclic context-free grammars including ambiguous grammars, while most of the LA parsing efficiency is preserved. The algorithm seems more efficient than any existing algorithms including the Cocke-Younger-Kasami algorithm and Earley's algorithm , as far a3 practical natural language parsing is concerned , due to utilization of LA parsing tables. The algorithm is an all -path parsing algorithm; it produces all possible parse trees (a parse forest) in an efficient representation called a "shared-packed forest. " This paper also shows th at Earley's torest representation has a defect and his algorithm cannot be used in natural language processing as an all -path parsing algorithm.