Web Page Classification using Iterative Cross-Training Algorithm

The paper presents a generalization of Iterative Cross-Training algorithm (ICT) which was previously applied to Thai Web pages identification [1]. The main concept of ICT is to iteratively train two sub-classifiers by using unlabeled examples in crossing manner. In this paper, we extend the algorithm in order to classify Web pages into course or non-course ones, which is a more challenging problem. We compare ICT against Supervised Naïve Bayes classifier and Co-Training classifier. The experimental results show that ICT still preserves its robustness on this new problem’s characteristic.