Optimization of Multi-Strategy Machine Translation System Based on AI Technology

As one of many methods of Artificial intelligence (AI), deep learning is inspired by the structure and function of human brain. By simulating the interconnection between neurons, an artificial neural network is constructed for repeated training until the expected value is output. Traditional English machine translation (MT) relies too much on expert summary, but there is a big error in English translation under weakening rules. In this paper, the semantic is Gaussian marginalized to obtain its rectangular window function and window feature vector, and the semantic correlation factors are added to the information entropy and information gain of the text. Compared with the traditional MT model, the evaluation accuracy of the proposed algorithm is improved by 25.68%. When English translation is based on AI technology, it is first necessary to translate the semantics of words. Mechanical intelligent recognition is used to select the coherent and key terms in semantics, and a certain language model is constructed to construct the English context and semantic connection. Combined with the fuzzy semantics of translation, the best translation solution is finally selected through discrimination and comparison.