The Fusion Learning On Technical Text Categorization Based On Decisiontree And SVM

With the online electronic information in the form of geometric growth, so endedJuly26,2008, Google search engine, number of pages indexed has reached one trillion. The flood of information from different industries, such as news, entertainment news, research papers, digital libraries, etc. In order to meet the needs of the rapid development of Internet, Many of the past in print journals have published papers will bemoved to its own publications on the Internet, especially e-journals in science and technology and the emergence of digital libraries has greatly enriched the knowledge resources of cyberspace.Electronic scientific literature examining how to achieve automaticacquisition of topic-oriented, automatic classification is a Web resource development and utilization of personalized service to achieve a meaningful subject, which is a very important part of the automatic text classification. This in-depth analysis of the decision tree and support vector machine (SVM) feature, the paper proposes a decision treeand support vector machines based on integrated learning classification of scientific literature, the main research work is as follows:First, the analysis of the decision tree learning method with SVM fusion researchsituation, the text of the segmentation algorithm, representation model and feature selection and other key Chinese text categorization theory and technology.Proposed based on support vector measure the importance of attribute and analyzed bycalculating the example of their relative to other feature reduction of the superiority ofstatistics.Classification based on support vector machine and support vector surface pointdistribution model, decision tree and support vector machine side surface of an effectiveapproach as the basic starting point to explore the integration of decision tree and SVM tolearn a new way. Focuses on the classification based on support vector points and theimportance of surface properties of shape characteristics and attributes measurement intervalpartitioning optimization methods, decision tree in order to achieve optimal performance.According to the characteristics of scientific literature, an algorithm based on fusion ofDecision Tree and SVM text classifier learning methods, based on this experimental study,and analyzed.