Based On The Research Of Web Bad Information Filtering System

With the rapid development of Internet technology, the network became animportant way we access information, network information and more important inpeople’s lives, but also the existence of network information of negative nature of alarge, large and mixed in, according to the information, filled with all kinds of badinformation, such as promoting superstition and violence information, reactionary andextremist messages and other illegal information and more. For these adverseinformation about the problem, we consider the network has become the main concernof information security issues.In general, the traditional data processing mainly by the involvement of staff,discussed the data supporting such an approach things with the data discussed in theslow work of human intervention in certain people. And when data is large, when thepossibility of human intervention is becoming increasingly smaller. So along with theDB technology, distributed computing, machine learning, artificial intelligence,technology, database technology has been widely used.Traditional filtering of information technology, key technology or mainly for the ipaddress filtering, these technologies can not be the ideal solution to the filtering ofundesirable information. This paper documents the traditional finishing technologies forin-depth material information discussed and put forward a number of feature selectionmethods, thus achieving a combination of machine learning and information filteringtechnology, documentation materials for adaptive filtering system.The main contents include the following areas:(1) information filtering technology, research status, and convenient overview ofthe actual significance, and elaborate on data mining and document classificationknowledge of materials.(2) documentation that discusses the theory of information filtering material.Material in the document during the study of information technology, materials thatdocument the basis for classification of the filter order. In the article discusses in detail the general method of classification of the document materials, including theclassification of the pre-order, feature selection, that relate to the classification modeland related algorithms. Further information on the current but also the main methodadopted by filtering and information filtering system composition and selection of themodels are also done in detail.(3) the use of filter material on the document several common characteristics in theselection method of exposition, a new feature selection method-multi-feature selectionmethod. Main work is: First, to test the results of comparing the various selectionmethods; the second discusses the impact of document classification performancematerials, the main factor; Third, feature selection methods that take more than theactual test results to be better than the single feature selection method.(4) The design of the adaptive information filtering system, the informationfiltering system is based on the vector space model associated with the traditionalinformation filtering system through the line on the adaptive capacity of the expansion.The system can follow the user’s feedback, automatically select the appropriate criticalvalue in order to achieve the purpose of providing filtered results. Evidence shows thatthe modified system, a significant increase in filter performance.