One-against-one fuzzy support vector machine text categorization classifier

The growth of the internet information delivery has made automatic text categorization essential. This investigation explores the challenges of multi-class text categorization using one-against-one fuzzy support vector machine with Reuter¿s news as the example data. While the fuzzy set theory is incorporated into the OAO-SVM in the classifying module, the influence of the samples with high uncertainty can be decreased as the fuzzy membership functions are to used to weigh the margin of each training vector. The performances of four different membership functions on one-against-one fuzzy support vector machine are measured using the macro-average performance indices. Analytical results indicate that the proposed method achieves a comparable or better performance than the one-against-one support vector machine.