A Practical Algorithm for Intersecting Weighted Context-free Grammars with Finite-State Automata

It is well known that context-free parsing can be seen as the intersection of a contextfree language with a regular language (or, equivalently, the intersection of a context-free grammar with a finite-state automaton). The present article provides a practical efficient way to compute this intersection by converting the grammar into a special finite-state automaton (the GLR(0)-automaton) which is subsequently intersected with the given finite-state automaton. As a byproduct, we present a generalisation of Tomita’s algorithm to recognize several inputs simultaneously.