Research And Design Of Intelligent Web Filtering System

Nowadays,Internet has become the part of human society survival and development.However,the opening leads some information such as eroticism,violence,evil religion etc is harmful not only to our heart but also to my country.As a result,how to filer the bad information and build up the health,safety information environment has become a research focus of the internet development in the future.Firstly,analyse traditional web technologies such as database filter(IP,URL),key words filter,content filter etc,and make a comparative analysis from filter speed and flexibility,the technical difficulties,deceptive,coverage scale in InternetThen,put forward neural network to the web filter system.In order to further optimize the ANN,proposed ANN training based on compound fitness simplex particle swarm optimization(CF-SPSO-ANN).The model use particle swarm optimization(SPSO) improved ANN learning algorithm,in order to improve optimal rate and the convergence speed of the network;The model use composite fitness(CF) improving ANN learning rule,in order to improve the network generalization ability.Experiments of IRIS data sets showed that CF-SPSO-ANN is better than CF-SPSO-ANN and CF-PSO-ANN in prediction accuracy and antinoise ability.Later,proposed method that by synthetically use of various above technologies to design a multi-layer web filter model:first layer is database filter(IP,URL),and second layer is key words filter,content filter.Finally,combined the thoughts of multilevel and intelligent,proposed an optimization model based on ANN multi-level smart web filtering model(AOM-ML-SWFS).Apply CF-SPSO-ANN to the illegal document filter modules.In order tO improve the system of intelligent,this model to log and filtration, reorganized training and learning library,formed learn- adjust-learn...Experiments show that this system has better reusability and expansibility, adaptability,in web filtering speed and precision,etc.This system can be mounted on a network information providers and users in the middle of the dedicated server,through the gateway or proxy monitor users and shield bad information.