Research, Key Technology For Information Filtering Based On Vector Space

With the rapid development of computer and communication technology in the world andthe popularization and application of Internet, more and more commercial and daily activities arecarried on through Internet. The network becomes closer with people's life. The informationfrom the Internet grows rapidly, which has brought the two-sided effect. On the one hand we canget much rich and latest information from the Internet, on the other hand we often feel quitehelpless when facing the ocean of information because the network information is vast and itscontent is numerous and disorder. Furthermore, the freedom of network information provides theconvenient place for the harmful information while providing the useful information. Weinevitably face a lot of harmful information when we surf on the Internet. So it arouses our moreand more attention to how to help people effectively choice and use the interested informationbut reject the information which is not related and harmful as far as possible and how tosupervise the network users especially young students when they surf on the Internet but notaffect the normal visit to the network.This thesis covers each processing stage of the information filtering and makes research andstudy on the following aspects with the two main indexes of filtering accuracy and speed ofinformation filtering model:1. Analyzing the current information filtering model and making further research on theinformation filtering based on the vector spaceThe thesis analyzes the current development process and tendency of information filteringand analyzes the essential technology and the related knowledge with which the informationfiltering model involved. It studies the detailed process of information filtering model based onthe vector space as well as the advantage displayed by this model. At the same time, in view ofthe present questions it discusses some links of the model which can be improved.2. Researching and the improving the technology of support vector machineIt studies the key part of the information filtering and introduces the support vector machineto the information filtering. On the question of harmful information filtering on Internet itimproves the technology of support vector machine on the base of analyzing the elementaryknowledge of support vector machine and the advantage in information filtering. Taking thefiltering content into consideration it proposes the idea of delamination of support vectormachine;taking the acquiring training knowledge into consideration it proposes the idea ofincremental study based on users feedback;taking the training material function intoconsideration it proposes the idea of fuzzy support vector machine.3. Proposing an information filtering plan with multilayer, multi-strategy and distributionaltypeThe thesis analyzes various characteristics of network information filtering especiallyharmful information filtering and proposes an information filtering plan with multilayer,multi-strategy and distributional type. First, this plan uses different technologies to filterseparately in different layers of network system. Second it combines the users' cooperation andcontent filtering to realize the multi-user cooperation by the means of users feedback and makeadjustment to content filtering profile by the feedback information. At last it separates the studythe process from the filtering process to further realize distributional processing and thus avoidethe system bottleneck.4. Using a new plan to design and realize an information filtering systemAccording to the idea of delamination and modularization design the thesis uses a new planto realize a now information filtering system. The new system has the characteristic of goodreusability, extensibility and adaptability. According to the experiment this system enhances therate of complete examination and accurate examination in the information filtering.