语义块主辅变换在机器翻译系统中的应用

Transformation of Main-Sub Semantic Chunks and Its Application in MT Systems

传统机器翻译系统的研究正在经历雪线困境(就是指机器翻译系统很难超越的译准率极限———70%).HNC理论针对传统机器翻译方法的弊端进行了深刻的反思,为突破雪线困境提出了一种全新的思路.语义块主辅变换是HNC机器翻译引擎中六个"过渡处理模块"的一部分.语义块主辅变换研究对于提高机器翻译系统的性能具有重要意义.本文简要介绍了HNC的机器翻译观,重点阐述语义块主辅变换的原理,并通过一个示例说明了语义块主辅变换研究如何来提高机器翻译系统的性能.

Traditional MT research has encountered the mountain-snow-line condition, which means the translation veracity of MT systems could not surpass the limit of 70%. HNC theory deeply reflects on the shortcomings of traditional MT research and puts forward a new way of breaking through the snow-line. Transformation of main-sub semantic chunks is one of the six kinds of transitional processing, which are kernels of the HNC MT engine. Transformation of main-sub semantic chunks plays an important role in advancing the MT systems. This paper briefly introduces MT perspectives of the HNC theory, explains principles of the HNC MT engine, and shows how the transformation of main-sub semantic chunks can be applied into advancing the performance of traditional MT systems.