Analysis of Web Objects Distribution

Understanding how the web objects of a website are demanded is relevant for the design and implementation of techniques that assure a good quality of service. Several authors have studied generic profiles for web access, concluding that they resemble a Zipf distribution, but further evidences were missing. This paper contributes with additional empirical evidences that confirm that a Zipf distribution is present in different domains and that its form has changed from past studies. More specifically, the α parameter has become higher than one, as a consequence that the popularity factor has become more critical than before. This analysis also considers the impact of web technologies on the characterization of web traffic.