Research And Application On Web Text Mining With SVM

Web has become a significant method of gaining information. Offering abundant information sources, vast web pages also put forward a challenge at the same time. It is how to get wanted information from the information ocean quickly. The technology of web text mining is a kind of solution. While, the application effect of web text mining technologies in existence is not ideal enough, and cannot fit the need of web documents expanding in high speed. So, developing new web text mining technologies has become a research hotspot. Direct against the above situation, this text has done the work of three following respects mainly.Firstly, this text introduces the basic concept, kind and method of web data mining, with the concrete procedure and correlated theories of web text mining. We introduce technologies of character denotation and character pick-up especially and bring forward a new character selection method. This method can select the characters having the most sort denotation meaning and benefiting to classifying most, with keeping down the character space greatly at the same time.Secondly, we study the Statistical Learning Theory (SLT) and Support Vector Machine (SVM) Theory seriously, and discuss classifying algorithm and kernel function. We expatiate the research and application status of Support Vector Marchine, and point out some important issues which is to be resloved when researchers do further research of SVM.Finally, we combine SVM with Incremental Study into web text mining and put forward a new SVM Incremental Study algorithm. It can abnegate useless samples and make the knowledge of study objects accumulate. This algorithm, in the incremental study question, is more effective than the traditional support vector machine, with assuring the classify accuracy.