Human-machine Translation Model based on Artificial Intelligence Translation

Abstract

The development of artificial intelligence (AI), cloud computing, Big data, and other technologies have promoted the rapid evolution of the data centre industry. As a new human-computer interaction mode, human-computer translation is essential in information transmission. However, many translation systems have certain defects, such as high error rates and low accuracy rates, and cannot meet personalized needs. This article, taking AI translation as the research object, analyzed attention mechanisms and other related technical methods, summarized and explored the limitations of traditional translation methods, and constructed a human-machine translation model based on AI translation, which had a clearer understanding of translation quality differences between various methods. It provides a reference for further improving translation models and exploring the application of AI in translation. The comprehensive automation and manual evaluation methods have proven that AI-based human-machine translation effectively improves the insufficient matching between text and context, enhances the accuracy of intelligent recognition and semantic expression, and sets a subjective evaluation score of 1-10 for users, with a score of 6 or above as qualified. The results showed that the language fluency score has increased from about 4.9667 for traditional Statistical Machine Translation (SMT) to about 6.6333 for AI-based human-computer translation. The AI-based human-computer translation model can improve the efficiency and accuracy of language input, enhance the correlation between intelligent recognition semantic features and text, and promote the development of AI recognition functions, thus providing users with accurate, efficient, and high-quality services and achieving automatic processing natural language understanding and output.