Generation of sentence parse trees using parts of speech

Abstract

This paper proposes a new corpus-based approach for deriving syntactic structures and generating parse trees of natural language sentences. The parts of speech (word categories) of words in the sentences play the key role for this purpose. The grammar formalism used is more general than most of the grammar induction methods proposed in the literature. The approach was tested for Turkish language using a corpus of more than 5,000 sentences and successful results were obtained. © 2004 Springer-Verlag.