Visual Area Classification for Article Identification in Web Documents

In the World Wide Web, the news and other articles are usually published in complex HTML documents containing many types of additional information that is not explicitly marked. In this paper, we propose a visual information analysis approach to the article discovery in complex HTML documents. We use a classification approach for the identification the important parts of the article within the page and we propose an algorithm for the detection of the article bounds within the page. Finally, we provide the results of an experimental evaluation.