Customizable instance-driven webpage filtering using graph-based semi-supervised active learning

Abstract

The World Wide Web has been growing rapidly in recent years, along with increasing needs for content-based webpage filtering. Most of the existing filtering systems, however, cannot easily satisfy the personalized filtering demands from different users at the same time. To address this issue, this paper presents a customizable instance-driven webpage filter strategy, which utilizes graph-based semi-supervised active learning. In the proposed strategy, a semi-supervised active learning approach is applied for obtaining a precise description of the webpage class based on the small-sized user instance set provided by user himself. Subsequently, a Bayes classifier is created over the enlarged training set. By this way, different webpage filters are produced by using users' own demands, so that the users are able to focus on their interested classes. Experimental results show the promise stability and high performance of our proposed method. © 2009 by Binary Information Press.