English writing big data intelligent correction system integrating attention mechanism algorithm model

With the rapid development of artificial intelligence technology, the demand for intelligent tools in the field of education is increasing. As an important means to improve learning efficiency, intelligent English writing correction system has attracted much attention. However, traditional manual correction faces challenges such as time-consuming and low accuracy (for example, it takes an average of 15 min for teachers to correct a single composition, and 73% of teachers think that grammar correction is the most time-consuming). In this paper, BiLSTM deep learning model integrating multi-head attention mechanism is proposed to build an intelligent correction system for English writing big data. Through hierarchical attention weight calculation (word level + sentence level) and multi-task learning optimization, the system is outstanding in grammar correction, semantic understanding and logical analysis. Experiments show that the F1 score of the system on the TOEFL test set is 91.5%, which is 4.8% points higher than that of the traditional BiLSTM model, and the average correction time is only 15.8ms/test, which provides an efficient auxiliary tool for teaching.