Research on English text summarization and machine translation technology based on machine learning

Machine translation technology uses machine learning methods, the translation model obtained is obtained after training on parallel corpora, and it is the mainstream method in current machine translation. The English text summary and translation system based on machine learning can help people obtain key information of English texts. In response to the problem of low translation accuracy, this article optimizes model parameters, model structure and data enhancement. It further improves the BLEU value by adjusting parameters such as batchsize during model training, alpha value and beamsearch value during prediction, reproducing Transformer multi-layer representation fusion, and performing data enhancement through "back translation". It uses the IBM model5 in moses for statistical machine translation, uses the Transformer in tensor2tensor for neural machine translation, and combines the unsupervised machine learning Textrank method for extracting text summaries to build a neural machine translation system, which can effectively extract key information from long texts, thereby further improving the effectiveness of translation.