Research On The Key Technologies Of Network Information Filtering Based On Optimize Model

With the development of information technology and the popularization of the Internet, the amount of information on the network has raised rapidly, the majority of computer users are seriously troubled by the information overload and the information pollution. The emergence and development of network information filtering provide a better choice for people to access to the information quickly, accurately and completely. Information filtering technology is a systematic approach. It matches the dynamic data stream to users'request and extracts the preference information to users. This paper presents a method that put emphasis on the key technologies of information filtering. The research is done as follows:1. Designed a simple and efficient Chinese word automatic segmentation method to apply to network information filtering systemConsidering Chinese word automatic segmentation is the basis of information filtering and it has a direct impact on the result of filtering, through analysising and researching on current primary Chinese word segmentation methods, we have designed a simple and efficient segmentation method to apply to high-speed network environment which has a large quantity of new words and rich language. The most advantage of this method is simple, high-speed and efficient in recognising new words. This method is also a reference for people to research and develop practical Chinese automatic segmentation systems.2. Provided a class-based weight computing and a single web text weight computing methodAt present, as many filtering systems do not consider the application in actual situation to distinguish the weight computing of features, this paper has provided two weight computing methods which are applied to two different processes. During the phase of training user profiles, we should apply class-based weight computing which adjusts feature importance on a text, a category, even a data set. During the filtering phase, at a moment only one document through the network, so we provide a single web text weight computing method, which adjusts weight mainly according to the structure of the text.3. We use the evolutionary mechanism of improved genetic algorithm to build and optimize user profiles and introduce gene expansion and incremental learning into genetic algorithm to update user profilesAs to user profile optimizing, most of the information filtering systems separated the building of profile from those updating. In this paper, taking the access and update of profile as a process of learning→adjusted→learning again→adjusted again……, after research and analysis comprehensively, we use the evolutionary mechanism of genetic algorithm to build and optimize user profiles. In order to make full use of the advantages of genetic algorithm and avoid the disadvantages of traditional genetic algorithm, this paper has carefully designed and improved some strategies of the genetic algorithm such as fitness function and genetic operators etc. The crossover and mutation operator which may result in the problem of premature convergence of the process and local minima has been improved. We also design adaptive crossover probability and adaptive mutation probability according to the process of evolution. Finally, gene expansion and incremental measures are introduced into genetic algorithm to update user profiles.4. We put the novel algorithm into action and design a network model of information filtering system of three strategies (URL-based filtering, keyword-based filtering and content-based filtering) and achieved satisfactory results.