Abstract
Web provides people a convenient way to disseminate and search information. Due to the growth of dynamic page generation techniques, the amount and the complexity of Web pages has been increasing explosively, as has the information contained within Web pages. Redundant information is distributed throughout a page, making it difficult to automatically identify the useful information in that page. In this paper, we propose and implement a simple Web importance extraction and labeling system based on the analysis on content information and vision information of a Web page. We apply the information theory on the document object model (DOM) trees of pages and extract the vision information for each block to evaluate their importance. Results show that our system effectively extracts and labeling the importance of a page and provides a powerful surfing interface for small display device browsing. Experiments on several Web sites show high performance to meet the users' information focus. © 2005 IEEE.