The Research On A Real Time Monitoring System For Bad Information Based On Artificial Neural Networks

Internet makes people easily obtain all kinds of information, however it also stimulates the promulgation of bad information and brings a lot of bad effect. So we need monitor and filter them. Based on the deep study on the current status, a real time monitoring system model for bad information based on artificial neural networks (ANN) is given in this paper and has been implemented by software.Firstly, this paper analyses the principle, the characteristic and the main application of ANN and places emphasis on the analysis of the BPNN and BP algorithm. To solve BPNN's shortcomings such as the slow training speed, inclined to immerse into the partial least point, bad generalization capability and hard to ensure the number of hidden layer, a special two-hidden-layer artificial neural network algorithm (STNNA) is proposed in this paper. After describing the principle and the major steps of this algorithm, the kernel of realization is given. Experimental results indicate that the generalization ability, training time and the architecture optimization of the networks have been improved obviously in this algorithm.Secondly, an intelligent decision-making problem for bad information is studied in this paper. After analyzing existing information filtering model and classifying algorithm, an intelligent decision-making system for bad information which is based on ANN is proposed. The system founds on the vector space model and adopts STNNA algorithm. As a result, it has some excellent characteristics such as rapid decision-making, high rate of accuracy and self-learning. Finally, combining the intelligent decision-making system for bad information and the technique of network sniffer, this paper proposes a real time monitoring system model for bad information and dissertates in detail the function and the implement of each module in the system model. Aiming at the characteristic of high network data stream, this paper proposes a solution for data overload. After tested, it shows that this system can solve several inter-contradiction problems that are confronted in the study of bad infor- mation inspection such as veracity, real-time and the feasibility in economy and has a good interface.