Unsupervised method for parsing coordinated base noun phrases

Abstract

Syntactic parsing is an important processing step for various language processing applications including Information Extraction, Question Answering, and Machine Translation. Parsing base Noun Phrases is one particular parsing issue that is not handled by current state-of-the-art syntactic parsers. In this paper we present research that investigates the base Noun Phrase parsing problem. We develop a base Noun Phrase parser based on several statistical models that provide promising results on a test set of 538 base Noun Phrases. The parameters of the models are estimated from the web in the form of web counts. This makes our method unsupervised with no training data being needed. © Springer-Verlag Berlin Heidelberg 2007.