This paper borrows the concept of entropy and refers the information entropy to the computer-aided English translation to build up the information entropy assessment and evaluation model and the information entropy-based English translation optimization model. The information entropy operation formula is used to calculate the fuzziness between the original text and the target translated text, and at the same time, the segmentation function is utilized to find the approximation value of the information entropy value obtained in a sufficiently fine segmentation of the translated text. Obtain the rectangular window function of semantic Gaussian marginalization, project the information entropy data, add the semantic correlation factors of words, identify the features of semantic information, and realize the optimization of English translation. Simulation experiments are carried out to verify the performance of the model. The information entropy evaluation and analysis model in this paper can achieve convergence after 45 iterations. In contrast, the word error rate and the cut-off error rate of the English translation optimization model are both lower than 10%, and the feature recognition and classification are higher than 80%. The fuzzy statements in Chinese writer Lao She’s work “Teahouse” are selected to explore the naturalness of translation. The information entropy value of the translation optimization model in this paper is closer to the original text, and the English translation’s naturalness is better than that of the GLR model.