Natural language parsing: A new characterization of attachment preferences
暂无分享,去创建一个
Abstract : Several authors have tried to model attachment preferences for structurally ambiguous sentences that cannot be disambiguated from semantic information. These models lack rigor and have been widely criticized. By starting from a precise choice of parsing model, it is possible to give a simple and rigorous description of Minimal Attachment and Right Association that avoids some of the problems of other models.