Existing question-answering systems often answer questions only through database search or manual question-answering methods, and there are certain problems with the accuracy and effectiveness of question-answering. In this paper, based on the semantic similarity related algorithm, we design an intelligent Q&A system for English based on semantic similarity calculation. Firstly, we modeled the lexical technology and educational text representation, proposed an improved TF-IDF weight calculation method to assign weights to professional and non-professional words respectively, and calculated the similarity between questions by using the calculation method of cosine similarity in the vector space model. Students’ classroom questioning data were collected to analyze the question-answering accuracy, question-prediction performance, and student application effect of the designed system.Five students used the intelligent module of the system to answer questions about the knowledge of Fundamentals of Grammar, and the checking accuracy rates were 91.43%, 95.83%, 96.55%, 92.86%, and 92.59%, respectively.With regard to the self-directed learning of the five English courses, the experimental classes’ learning hours were far higher than that of the control class, and the frequency of questioning is also higher and the attitude is more positive. This study applies artificial intelligence technology to achieve more accurate and efficient Q&A in English teaching, which provides new ideas and methods for English teaching innovation.