Research Of WEB Mining Applying In The Monitoring System Of The Network Policing Information

Faced with the complicated situation on the crimes and growing tremendous information on the network policing information, it is impossible to regulate them comprehensively by collecting and browsing solely, with the disadvantage of losing useful information and untimely cracking the cases.For these problems, with the means of the academic analysis and experimental research, the paper emphasizes the arithmetic and matrix of these pivotal technologies, designs a monitoring system of the network policing information, advances a discovery method of one certain WEB texts mining, and makes clustering analysis about the network policing information on the basis of texts similarity, in order to learn the immanent relations among the policing information and support the network real time supervision technically for the public security organs in our nation. The achievements of the paper are as follows:First and foremost,this paper designs a network monitoring systematic model on the basis of multi Agent system according to the practical needs in the public security work. And the functions of Agent will be presented in detail;Then the paper is to introduce some critical techniques, including information collecting, information cleaning and Chinese segmentation. Furthermore, the paper will present the design and realization of some certain data bases of text mining models which show the information about the crimes, and the means of improving data extraction through text classification after the formation of data bases; Thirdly,this paper brings forward two improvements about the traditional K-means algorithm, by means of analyzing the characteristics and shortcomings in the traditional K-means algorithm and the existing improved algorithm. Firstly, This paper brings forward the method of sector segmentation.After being classified into vertical and lateral steps crossly, the final assorting k is attained as the value of k , the initial clustering result of K-means algorithm; Secondly, This paper brings forward the K-means algorithm base on a verage similarity of text algorithm;Last but not the least, This paper designs a subsystem of text clustering. After being put into the practice, the improved K-means algorithm is efficient.