The Research Of Web Text Mining Based On Improved SVM

With the rapid development of Internet, the information explosion is coming. Because of the open and dynamics of Internet, it's difficult to user to get information quickly and exactly that they need. So how to get the valuable information from internet has been current research hot spot. And Web text mining technology has been a method of solving these problems. Web text mining uses these technology including data mining, machine learning, natural language processing, information retrieval and knowledge management to process and analyse non-structured or semi-structured text and obtains the valuable knowledge.There are many text categorization methods at present such as the Nearest Neighbor method, Bayesian Networks decision trees, neural networks, support vector machines, vector space model, regression model, etc. By analyzing the disadvantage of present text categorization method, the paper proposals improved web text categorization method based SVM.Firstly, this text introduces the basic concept, kind and method of web data raining, with the concrete procedure and correlated theories of web text mining. Secondly, we study the Statistical Learning Theory (SLT) and Support Vector Machine (SVM) Theory seriously. We explain the research and application status of Support Vector Machine and point out some important issues which is to be resolved when researchers do further research of SVM. Finally, we combine SVM with active learning, introduces an improved active learning SVM methods used to web text categorization. Compared with the general SVM,it can reduce the number of examples effectively on the premise of keeping correctness of the classifier.