Efficient Parsing Using Recursive Transition Networks with Output

* Final gross prices may vary according to local VAT.

Get Access

Abstract

We describe here two efficient parsing algorithms for natural language texts based on an extension of recursive transition networks (RTN) called recursive transition networks with string output (RTNSO). RTNSO-based grammars can be semiautomatically built from samples of a manually built syntactic lexicon. Efficient parsing algorithms are needed to minimize the temporal cost associated to the size of the resulting networks. We focus our algorithms on the RTNSO formalism due to its simplicity which facilitates the manual construction and maintenance of RTNSO-based linguistic data as well as their exploitation.