Model Of Research And Practice, For The Information Filtering Based On Self-learning Mechanism

With the rapid development of NII based on the Internet, information technology hasbeen walked into every aspect in our life. The information in the Internet rises in exponentialway. The increase of information has two effects, one is it makes people get abundant andrecent information easily, the other is some information such as eroticism,violence,evilreligion etc is harmful to our heart. As a result, how to filer the information irrelevant to ourdemand and pick up the information we need is getting the hotspot in the research of theInternet.This paper mainly studies the lawless information filtering, the contents include everyphase of information filtering, the following questions are the research keystone around thetwo aspects, precision and velocity.1,The analysis of information filtering model in existence and the application of machineleaning in these models.Firstly, analyzes the development process and development trend of information filteringmodel, second, analyzes the critical technique and the relevant knowledge, Thirdly, points outthe defaults in the current information filtering system, such as the deficiency of self-leaningand self-adjust in constant varying environment, at last, discusses which phase in the processof information filtering can be optimized by machine leaning.2,This paper puts forward the method of creating the stop-word table based on statisticsand the optical algorithm of features pick-up based on genetic algorithm.Splitting words and feature picking up are two important sides in information filteringmodel. First, this paper analyzes the characteristic of stop-word, second puts forward amethod of generating stop-word table based on feedback, this method improve the precisionof splitting words. Additionally, to improve the precision to feature picking up, this paperconstructs two level character database and optimizes the central character database bygenetic algorithm and gives a new feature picking up optical algorithm.3,A new information filtering model based on self studying mechanism is put forward.The core technique of information filtering includes splitting words, feature picking up,document denotation and document classifying. There is many problems in informationfiltering, such as character item's weight has less capability of Stat., difficulty of ascertainingthe number of character items, less relation between character item's weight and the valvevalue of classifying algorithm. As a result of the upper problems, this paper studies how touse Racchio method,decision-making tree method,leaning method based on example,ANNmethod etc to optimize the information filtering model.