Text content security detection method based on deep learning

The invention relates to a text content security detection method based on deep learning. The method comprises an algorithm part, and a corresponding detector is a convolutional neural network framework. The overall architecture of the algorithm adopts a long short term memory (LSTM) network, and the structure is a recurrent neural network, is simple to implement and has a long-term memory function; on the basis, an attention mechanism is further added, limited resources of a computer can be efficiently utilized through the mechanism, parallelism can be achieved during use, and the accuracy can be improved under the condition that network parameters are reduced. Besides, a residual structure is added to the algorithm part, so that a learning result is more sensitive to fluctuation change of network weight, and meanwhile, a residual result is more sensitive to fluctuation of data. According to the method, a depth-based method is provided, different text content detectors can be constructed by changing model pa