权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

Studies on Corpus Creation and Use for Linguistic Research

语言学研究的语料库创建和使用研究

基本信息

批准号：
15300046
负责人：
MATSUMOTO Yuji
金额：
$ 9.28万
依托单位：
Nara Institute of Science and Technology
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (B)
财政年份：
2003
资助国家：
日本
起止时间：
2003 至 2005
项目状态：
已结题

项目摘要

As for the research for language processing, we augmented the language analysis tools we have been developing, such as Japanese morphological analyzer and Japanese dependency analyzer, for Chinese analysis.As for development of dictionaries, we implemented unknown word analysis system for Chinese, and extracted candidates of new word entries by running the system on a large scale Chinese corpus. Through this experiment, we could successfully construct a large scale Chinese dictionary with about a hundred thousand word entries. For Japanese, we described the constituent word information of Japanese compound words and registered these information in the dictionary. For English, we developed a method for distinguishing literal and idiomatic uses of English multi-word expressions, and showed a high accuracy in distinguishing them.As for the corpus tool development, we made a detailed design of the database schemes for annotated corpus and dictionary entries, and re-implemented the corpus management tool based on these schemes. We also implemented the error correction functions for part-of-speech and dependency analysis errors and designed and implemented the interface for the functions. The visualization function for showing phrasal chunks and their dependency relation, on which one of the error correction functions is realized.The developed corpus management tools are made open to public and we hold two seminars to make it open and to explain the usage to those interested in using the system, aiming at collecting the feedback from the users. We also opened a Web page for introducing and downloading the tools.

在语言处理的研究方面，我们扩充了我们一直在开发的语言分析工具，如日语词素分析器、日语依存分析器等，用于中文分析。在词典的开发方面，我们实现了汉语生词分析系统，并通过在大规模中文语料库上运行该系统来提取候选生词。通过这个实验，我们成功构建了一个包含十万词条的大型汉语词典。对于日语，我们描述了日语复合词的构成词信息，并将这些信息登记在词典中。对于英语，我们开发了一种区分英语多词表达的字面用法和惯用用法的方法，并表现出了较高的区分准确率。在语料库工具开发方面，我们对标注语料库和词典词条的数据库方案进行了详细设计，并基于这些方案重新实现了语料库管理工具。我们还实现了词性和依存分析错误的纠错函数，并设计和实现了该函数的接口。展示短语块及其依存关系的可视化功能，在此基础上实现纠错功能之一。开发的语料库管理工具已向公众开放，我们举办了两次研讨会，向有兴趣使用该系统的人开放并解释其使用方法，旨在收集用户的反馈。我们还开通了一个网页来介绍和下载这些工具。