北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2004, Vol. 27 ›› Issue (4): 1-12.

• •    下一篇

自然语言理解的全信息方法论

钟义信   

  1. 北京邮电大学 智能科学技术研究中心, 北京 100876
  • 收稿日期:2003-04-21 出版日期:2004-04-28
  • 作者简介:钟义信(1940—), 男, 教授, 博士生导师. E-mail:yxzhong@ieee.org

Comprehensive Information Based Methodology for
Natural Language Understanding

ZHONG Yi-xin   

  1. Center of Intelligence Science and Technology Research, Beijing University of Posts and
    Telecommunications, Beijing 100876, China
  • Received:2003-04-21 Online:2004-04-28

摘要: 在经济全球化需求的推动下,世界在酝酿一场“自然语言信息技术革命” ,它的基础和核心是“自然语言理解”的理论与方法. “全信息自然语言理解方法论”是作者的“全信息理论”在自然语言理解领域的应用. 与文献中已有的其他工作不同,其主要特色是: 一方面,试图实现语法信息、语义信息、语用信息的综合利用;另一方面寻求“规则方法”和“统计方法”的和谐互补,从而有效增强对自然语言的理解能力. 近几年来,应用这一方法论完成了一系列自然语言理解方面的课题,取得了一批可喜的研究成果,表明全信息自然语言理解方法论具有很好的前景.

关键词: 自然语言理解, 全信息理论, 规则方法, 统计方法, 全信息自然语言理解方法论

Abstract: The “Comprehensive Information Based Methodology forNatural Language Understanding” presented in the paper is an application of the Comprehensive Information(CI) Theory, which was established by the author of the paper years ago formeeting the needs in artificial intelligence research. Different from other works in the field of natural language processing, the major features of the methodology presented in the paper include two folds. Firstly, instead of the syntacticor/and semantic information is used, all components of the comprehensive information are fully utilized for natural language understanding. Secondly, instead of using one or another approach preferably, both the rule-based approach and the statistics-based approach have been complementarily integrated. The successesof the research projects guided by the methodology and completed last few yearsprove the methodology promising.

Key words: natural language understanding, comprehensive information theory, rule-based approach, statistical approach, CI-basedmethodology for natural language understanding

中图分类号: