Computing and Informatics

To learn a Web browsing behavior model, a large amount of labelled data must be available beforehand. However, very often the labelled data is limited and expensive to generate, since labelling typically requires human expertise. It could be even worse when we want to train personalized model. This paper proposes to train a personalized Web browsing behavior model by semi-supervised learning. The preliminary result based on the data from our user study shows that semi-supervised learning performs fairly well even though there are very few labelled data we can obtain from the specific user.