Does anything like this exist for XML documents? Wonder if it could be used for identifying interesting information in web pages.