AdaBoost ranking results improvement by pairwise classifiers for web page classification

Abstract

The article concerns the analysis of information describing the web pages. The aim of the analysis is to support the process of their classification. Pages belonging to the specific class are characterized by the similar 'style' in terms of the form or the type of content presentation. Various characteristics are taken into account including inter alia, structural, visual, text, web and links features. During the construction of classifiers the AdaBoost algorithm was applied to create ranking list of classifiers. Then the pairwise classifiers were used to improve final classification. The paper presents the implementation of this solution and the results of experiments. © 2011 Springer-Verlag Berlin Heidelberg.