Cross-site scripting attack detection method and system based on deep learning

The invention discloses a cross-site scripting attack detection method and system based on deep learning. The method comprises the following steps: S1, acquiring XSS sample data in multiple channels; s2, performing sample construction based on the acquired XSS sample data and normal sample data, and preprocessing the constructed sample to generate a sample data set; s3, based on a deep learning method, establishing an improved model which connects a CNN model and a BiLSTM model in series and combines an attention mechanism, and extracting sample data set features; s4, utilizing a feature vector obtained by carrying out transfer learning through BERT to accelerate convergence of the model, and improving the detection efficiency; and S5, deploying an improved model which connects the CNN model and the BiLSTM model in series and combines an attention mechanism, collecting data, and carrying out cross-site scripting attack detection. According to the scheme, the accuracy rate, the recall rate and the precision ra