基于大规模语料库的多引擎语言翻译模型的构建

The Construction of Large Scale Corpus-Based Multi-Engine MT Model

提出了基于大规模语料库的多引擎翻译系统模型的构建,该模型主要包括三大部分:基于层次化长句分解和支持向量机基本名词识别的句法分析器、基于实例匹配的翻译子引擎、基于语块的统计翻译子引擎.该语言翻译模型综合各个翻译模型的优点,最大程度地提高机器翻译的准确率和召回率.实验表明该系统的各项指标都比较理想,尤其是处理效率很高.

We put forward a construction model of the large scale corpus-based multi-engine MT system,which includes three parts： the syntax parser based on hierarchical sentence decomposition and basic noun recognition of SVM（support vector machines）,the EBMT（example-based machine translation） sub-engine,and the chunk-based STMT（statistical machine translation） sub-engine.The multi-engine model covers all the advantages of various translation methods and improves the accurateness and recalling rate to the largest degree.Experiments show that the system performs very well in various indexes,especially in improving the processing efficiency.