文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > 高等教育 > 文学 > design of Chinese HPSG framework for data-driven parsing
侵权投诉

design of Chinese HPSG framework for data-driven parsing

Design of Chinese HPSG Framework for Data-Driven Parsing

aa Xiangli Wang, Shunya Iwasawa, Yusuke Miyao, Takuya Matsuzaki, aa,b,c Kun Yu and Jun’ichi Tsujii

Department of Computer Science, University of Tokyo,

Hongo 7-3-1, Bunkyo-ku, Tokyo, 113-0033, Japan

{xiangli, iwasawa, yusuke, matuzaki, kunyu, tsujii}@is.s.u-tokyo.ac.jp b School of Computer Science, University of Manchester c National Center for Text Mining aaa

Abstract. Data-driven parsing has been a main method for analyzing natural languages. We aim at exploring a data-driven Chinese parser, by basing it on Head-driven Phrase Structure the basis for a practical In this paper, 1) we present a Chinese syntactic structure system and 2) we design a primary Chinese HPSG framework.

Keywords: HPSG, data-driven parsing, Chinese HPSG framework, coverage, consistency. 1 Introduction

parser. It can deliver a parser with broad-coverage and high-accuracy. Some English data-driven syntactic parsers have been developed in the past (Charniak and Johnson, 2005; McDonald and Pereira, 2006; Miyao and Tsujii, 2005). Following the success of the research on English data-driven parsing, the same methodology has been applied to Chinese parsing (Levy and Manning, 2003; Wang et al., 2005; Guo et al., 2007).

the HPSG framework has been developed by our group and Tsujii, we follow the into an HPSG treebank, based on which we can obtain a large lexicon and a statistical model for Since the HPSG framework for English has been studied comprehensively (Sag et al., 2003), we can rely on the framework developed by linguists to establish the initial framework for a data-driven HPSG parser. For Chinese, however, although some linguistic studies have been conducted in the HPSG framework (Gao, 2000; Wang and Liu, 2007), they have narrowly focused on very specific phenomena, and a HPSG framework for Chinese which is comprehensive and systematic enough to cover wide range of phenomena in Chinese still does not exist.

As the first step towards a data-driven HPSG parser for Chinese, our work aims to design a Chinese HPSG framework, by which we can trigger lexical acquisition from an annotated corpus. enough to avoid superfluous ambiguities. To achieve these two goals, the framework should have a disciplined view of syntactic structures of the language and provide a simple but versatile data structure to treat diverse phenomena. Our experience of developing a data-driven parser for English

Copyright 2009 by Xiangli Wang, Shunya Iwasawa, Yusuke Miyao, Takuya Matsuzaki, Kun Yu, and Jun’ichi Tsujii

23rdPaci cAsiaConferenceonLanguage,InformationandComputation,pages835–842

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email