A Hybrid Approach to Error Reduction of Support Vector Machines in Document Classification

In this paper, we present a hybrid method of support vector machine and k-nearest neighbor to improve the performance of automatic text classification. The proposed methods first classify a given document using SVM which shows the best performance in text classification, and then is reinforced by k-NN for the documents that are not confidently classified by SVM. According to the experimental results, the hybrid method achieves the F-score of 85.2, which implies that the hybrid method outperforms SVM alone