Efficient parsing algorithms for general context-free parsers

Abstract A family of parsing algorithms for general context-free grammars is described. Its members have a top-down structure, while performing tests similar to those introduced in both LR and LL algorithms. A theoretical study shows that the algorithms have the same time and space bounds as Earley's algorithms, which are particular members of the family. Empirical comparisons show the effectiveness of the new members of the family, which appear to be definitely better than Earley's algorithms, except for a few pathological grammars.