Third Party Certification of HTTP Service Access Statistics

Abstract

Online advertisement represents a significant portion of Internet-related revenues. However, no reliable and widespread system for metering Web site popularity has emerged. As a consequence, the market is dominated by the few very well-known sites, such as the main newspapers and the most effective search engines, because their popularity is taken as a given fact. For many sites of medium to high popularity there is a great potential for an increase of advertisement profits, but third party guaranties for their declared access rates is needed. There has been surprisingly little work on Web metering, and most of this work is not concerned with certified statistics. In fact, Web counting is a difficult and complex problem, raising many issues of different kinds. Even when the institution running the Web server is trusted, the obtained statistics may not correspond to real usage. Some traffic is generated from within the local area network, mainly for the site’s development and testing, and should be eliminated. Some other traffic is generated by Web spiders and other automated search engines. This should also be treated differently. A large number of external user accesses are hidden by proxies at different network locations, and it would be interesting to quantify the number of such additional hits. An in depth analysis of these issues is found in [6].

The techniques described herein are subject to patent protection (clearance request sent to the Italian Patent and Trademarks Office, November 1998).