SVM-based Hypertext Information Categorization

An algorithm of hypertext information categorization based on SVM was introduced. The algorithm divided a large data-set into many non-intersecting subsets during training period, in which the samples are trained according to the batch and then many classifications are constructed. The classifications are optimized by error correcting output codes (ECOC), which reduced the amount of document to be studied in the deep level training phase. The application of this algorithm to hypertext information categorization underlines the theoretical results.