| With the rapid development of Web applications,the “we media" platforms based on Web services have become the gathering and cradle place of public opinion.In order to control the trend of network public opinion,government departments need to construct a set of effective public opinion guidance mechanism,and Web behavior simulation technology has become the hotspot technology to solve the problem.In addition,the technology has been applied widely in network security,Web service performance testing and so on.While the existing simulation technologies have the problem of simulated distortion in terms of traffic and time characteristics,user behavior patterns.To address this issue,the paper proposes a multi-level Web behavior simulation method in which the behaviors of Web session layer and user layer are simulated respectively on the basis of behavioral analysis.At the session layer,the interaction behaviors of Web services are simulated based on the analysis of the interaction behaviors of Web session.Firstly,the interactive process of Web services is described and analyzed.Then,on the basis of analysis,a simulation method of Web service interactive behavior based on state transition is proposed.Finally,a distributed parallel execution scheme of simulation is designed.The experimental results show that the simulation method realizes the automation and efficient simulation of Web service interaction process.At the user layer,realistic simulation of user behavior is realized based on the analysis of user mouse dynamics.In the stage of feature analysis,a user classification model based on mouse dynamics is presented to describe the differences of behavior patterns among individual users and to be used as an evaluation criterion for subsequent simulation effects.A weight-based voting algorithm is adopted in the model,and the conventional statistical features with newly defined individual features of the operations are used to enhance the classification performance of mouse dynamics.An accuracy of 84% with data of 90 s on a dataset of 20 users is obtained in the identification experiment of multi-classification.In the authentication experiment of two-classification,a false rejection rate of 6% and a false acceptance rate of 11% are gained.In the stage of behavior simulation,a simulation method of mouse operations based on template library matching is proposed,and the simulation scheme is improved in order to adapt to different simulation machines.The simulation results are evaluated with the user classification model proposed,showing that both two simulation schemes can achieve realistic simulation of multiple user behaviors.A prototype system of public opinion guidance i s designed and implemented on the basis of the above researches in which an efficient network army fighting function is realized based on the session layer behavior simulation,and the network army rearing function is achieved based on the user layer behavior simulation to improve the quality of network army fighting,enhance the level of public opinion guidance and the effect of confrontation.The system is tested in the actual network,and the results indicate that the system can be used by the relevant government departments for the actual network public opinion guidance. |
PDF Full Text Request