Methods of Improving Japanese-Chinese Machine Translation System through Machine Learning and Human-Computer Interaction

Abstract

With globalization, the exchange of people between different countries is becoming more frequent. Due to different languages, there are serious obstacles to personnel exchanges. To a large extent, they hinder the growth of industries such as economy, culture and tourism in each country. The emergence of Machine Translation (MT) has effectively improved the problem of language barriers, and greatly reduced the workload of translators in text translation. However, MT does not have the same flexible flexibility as human translation. It just translates the text word by word, which is often difficult to meet people's higher needs. This paper proposed to build a Japanese-Chinese MT system and integrate machine learning and Human-Computer Interaction (HCI) technology into the system. To further enhance the efficiency of the system, enhancement algorithms were also applied to the system to optimize the performance of the system. From the experimental results, in terms of BLEU (Bilingual Evaluation Understudy) index, the average BLEU index of the algorithm in this paper was 8.59, and that of the traditional algorithm was 6.55. In terms of translation precision, the average precision of the algorithm in this paper was 91.53%, while that of the traditional algorithm was 87.28%. In terms of translation readability, the average readability of the algorithm in this paper was 93.32%, while that of the traditional algorithm was 89.22%. By comparison, the average BLEU index of the algorithm in this paper has increased by 2.04; the average accuracy of translation increased by 4.25%; the average readability increased by 4.1%. From the above data, it was evident that the enhancement algorithm can optimize the performance of the Japanese-Chinese MT system well.