Query-Based Discovering of Popular Changes in WWW

This paper presents the method for retrieving and summarizing changes in topics from online resources. Users often want to know what are the major changes in their areas of interest. Usually, change detection applications are based on predetermined sets of web pages. User needs to provide the addresses of web pages in order to receive recent information about occurring changes. Our approach involves creation of dynamic web collection for a given area of user’s interest. Such collection would contain informative and up-to-date resources. Periodically, we monitor the set of pages in search for new textual data and detect significant terms to extract sentences reflecting popular changes within every period. Since many web pages can be static over long time, we propose a method for evaluating how up-to-date a web page is in context of a given topic. Each WWW page is scored according to the frequency and contents of its changes. The most valuable pages form a base for next change summarizations. Additionally we expand the web collection to include new, valuable resources by finding pages, which have similar characteristics to the top-scored pages from the collection.