On coordination disambiguation in a generative parsing model, with memory-based techniques for parameter estimation.

Abstract

This thesis is concerned with improving existing generative history-based probability models' treatment of noun phrase coordination and the development of memory based techniques for model parameter estimation. Lexicalised generative history-based parsing models have proven to be highly successful at robust and accurate parsing. Although such models already achieve impressive overall accuracy results there is nevertheless potential for improvement, particularly in areas of difficulty for such parsers, such as coordination disambiguation and, more generally, parameter estimation from sparse data.