基于马尔科夫树时态标注算法的自动机器翻译系统研究

Research on automatic machine translation system based on Markov tree temporal annotation algorithm

在汉英文化交流日益频繁的背景下,精确的汉英机器翻译成为一个重要议题。针对英语时态不一致的问题,研究提出了一种马尔科夫树时态标注算法,在这基础上结合了深度学习的Transformer模型,最终构建了自动机器翻译系统。结果显示,在包含有时态和无时态数据的对比中,时态标注的总准确率从0.670提升至0.720,动词准确率从0.676提升至0.725。此外,对于新词汇的标注准确率,双元结构从64.9%提高至84.9%,而三元结构从61.2%提升至92.9%。此外,结果表明,马尔科夫树时态标注算法的自动翻译系统,与未使用时态标注的基线模型相比,具有较高的翻译准确率。该研究对提高机器翻译的精度和可靠性具有重要意义,为机器翻译技术的进一步发展提供了有价值的方向。

In the context of increasingly frequent cultural exchanges between Chinese and English,accurate Chinese-English machine translation has become an important issue.To solve the problem of inconsistent English tenses,a Markov tree tenses annotation algorithm is proposed,and based on this,an automatic machine translation system is constructed by combining the deep learning Transformer model.The results showed that in the comparison of temporal and non-temporal data,the overall accuracy of temporal annotation increased from 0.670 to 0.720,and the accuracy of verb increased from 0.676 to 0.725.In addition,the tagging accuracy of new words increased from 64.9%to 84.9%for binary structures and from 61.2%to 92.9%for ternary structures.In addition,the experimental results show that the automatic translation system combined with temporal annotation has a higher translation accuracy than the baseline model without temporal annotation.This research is of great significance to improve the accuracy and reliability of machine translation,and provides a valuable direction for the further development of machine translation technology.