Web Page Classification Using Relational Learning Algorithm and Unlabeled Data

Applying relational tri-training (R-tri-training for short) to web page classification is investigated in this paper. R-tri-training, as a new relational semi-supervised learning algorithm, is well suitable for learning in web page classification. The semi-supervised component of R-tritraining allows it to exploit unlabeled web pages to enhance the learning performance effectively. In addition, the relational component of R-tri-training is able to describe how the neighboring web pages are related to each other by hyperlinks. Experiments on Web-Kb dataset show that: 1) a large amount of unlabeled web pages (the unlabeled data) can be used by R-tri-training to enhance the performance of the learned hypothesis; 2) the performance of R-tri-training is better than the other algorithms compared with it.