Automatic Extraction of Grammars From Annotated Text

摘要

The primary objective of this project is to develop a robust, high-performance parser for English by automatically extracting a grammar from an annotated corpus of bracketed sentences, called the Treebank. The project is a collaboration between the IBM Continuous Speech Recognition Group and the University of Pennsylvania Department of Computer Sciences. Our initial focus is the domain of computer manuals with a vocabulary of 3000 words. We use a Treebank that was developed jointly by IBM and the University of Lancaster, England.rnIn this past year, we have demonstrated that our automatically built parser produces parses without crossing brackets for 78% of a blind test set. This improves on the 69% that our manually built grammar-based parser produces.