The Application Of Active Defense And Audit Based On Large Data Analysis Of Application Log

With the advent of the era of big data,the government,enterprises and institutions have been increasing their investment in information technology,especially in recent years,the state has advocated the information goal of "large integration,high sharing and deep application".The categories and amount of government data resources have increased sharply,the concentration and sensitivity of personal data have further increased,the utilization and external sharing of data have become more complex,and data theft and ultra vires have occurred Querying data and so on cause the government to have the personal data information to divulge,the violation citizen personal privacy phenomenon to occur unceasingly,affects the government information security seriously,urgently needs to start from the new technical means such as machine learning,further strengthens to the government data resources protection.In recent years,the leak of personal information has become a hot media issue,caused various kinds of hype,caused adverse social impact,seriously affected the image of government organs,and damaged the achievements of government information development.Therefore,this paper would intend to study how to predict abnormal users' behavior more effectively based on neural network and decision tree through the application log data of one month.Firstly,study the basic of artificial neural network and decision tree algorithm,with emphasis on the characteristics and working principles of artificial neural network,as well as the algorithm of decision tree and the application scenarios.Secondly,extract the government log data in certain month for pre-processing and feature analysis,and establish an abnormal audit model on users' behavior based on artificial neural network and decision tree.By designing and optimizing the training data of the two models,the prediction model is gradually optimized and verified.Thirdly,predict the historical data according to the decision tree model and the neural network model.It is known that the neural network model has obvious advantages in the goodness of model fitting,the ability to simulate the initial data and the ability to predict the new data.Discuss the nonlinear mapping relationship of the deep neural network model and the nonlinear function movement.The ability of state processing,self-organizing self-adaptation and self-learning,and the effect of variable processing are all due to decision tree.