This paper presents a solution to keep available up-to-date information in a search engine whose scope is the content available within news web sites. This solution is based on the use of non-uniform policy to update the documents belonging to this scope. In order to use the non-uniform policy, we identify the most and the least recently updated documents, based on the idea in which it is supposed that the closest documents of the root of a site are the most modified ones. This hypothesis was verified through cm experiment within news sites. In order to demonstrate the efficiency of our solution regarding a traditional one, we performed a case study whose results showed that: our solution spent less time to make the new information available, it made fewer requests to the web server, it kept a high freshness of the scope and, finally, it kept the search engine index up-to-date for a much longer time than the traditional solution.
展开▼