SC文法功能体系

A NEW CONTEXT-SENSITIVE SUBCATEGORY (SC) GRAMMAR FOR MACHINE TRANSLATION

文法体系的研究一直是自然语言处理研究的核心问题之一.但是,由于自然语言本身所固有的复杂性和非规范性,多义问题始终未能得到圆满的解决.本文提出了SC文法(A SubCategory grammar for integrating Se-mantic and Case analysis),它是一种基于传统的上下文无关文法、语义文法,以及超前与反馈分析技术和格框架约束分析等技术的上下文相关处理文法.该文法不仅能继承传统的上下文无关文法的表示简洁、处理方便的特点,而且能实现语法和语义一体化分析和处理上下文相关以及复杂多义问题.

In this paper, we propose a new grammar for machine translation based on the technology of Hierarchical Compatible Unification, Forward and Feedback analysis as well as Backtracking Inference, etc. The grammar integrates the alvantage of simplicity of phrase structure grammar, the powerful context sensitive analysis of case grammar and the application of detail semantic knowledge of semantic grammar into an identical Subcategory Grammar System.