Research On Extraction And Evaluation Method Of Campus Network Security Situation Elements Based On Deep Learning

With the continuous advancement of information construction in colleges and universities,more and more online applications and services have penetrated into the daily operation of the campus,and the importance of campus network security has become increasingly prominent.The development of campus network because of the existence of network security facilities equipped with insufficient,equipment management and other security risks,threatening the stability and security of campus network.As a comprehensive security protection measure,network security situational awareness can evaluate the campus network security on the whole,check the gaps and prevent risks.Therefore,this paper carries out an in-depth study on the extraction and evaluation of campus network security situation elements,and the main work is as follows.1.In order to solve the problems of difficult security risk assessment and incomplete indicators in the campus network,this paper proposes the campus network security situation indicator system.The index system is based on the current network security situation assessment index system,combined with the characteristics of the campus network itself,the campus network security risk into the index system,provides a basis for the campus network security situation elements extraction and evaluation evaluation.2.In order to solve the problems of single measurement dimension,incomplete feature extraction of multi-source data,and ignoring the hidden relationship between data packets in the extraction of network security situation elements,this paper proposes a method combining CNN and BILSTM to extract situation elements.This method extracts the temporal and spatial features of the data from two aspects of time and space,and mines the hidden relationship between the data at the same time,so as to extract the data features more comprehensively and improve the extraction effect of situation elements.3.In order to solve the problems of unstable multi-source data processing and gradient descent of RNN,this paper proposes the Bigru-Attention model.At the same time,the network structure and parameters of the Bigru-Attention model were designed in detail to improve the accuracy of the evaluation.Then,the campus network data was used for verification.Through the comparison of several groups of experiments,it is found that the method combining CNN and BILSTM to extract network security situation elements and the method using Bigru-Attention model to evaluate campus network security has certain advantages,and the results are more accurate.