With the acceleration of economic globalization, the importance of translation services has gradually manifested. The machine translation technology has a long and glorious history from statistical machine translation to neural machine translation, At present, machine translation software is emerging, mostly following the rules of natural language processing. In order to achieve quality and efficiency in translation, the various ambiguities in the language should be processed accurately and the speed of translation should be increased, while the anthropomorphism of machine translation should be enhanced and the machine's knowledge of human language should be constantly improved. Therefore, it is important to strengthen the research of machine learning technology, and to insist on the research on machine analogy translation technology based on machine learning is an important research topic.