90 likes | 117 Views
This analysis by Peter Kamm delves into the access methods of the blogosphere community, examining traffic characteristics and communication patterns from three perspectives: Server View, User View, and Object View. The study covers Pareto distribution, diurnal bursty patterns, blog popularity trends, and different blog types. With data spanning over a terabyte, it includes extensive information on server requests, excluding crawlers and errors. The research sheds light on the impact of search engines, power law relationships, and the role of social networks in blogosphere patterns.
E N D
A brilliant and insightful analysis of the access methods of the blogosphere community Peter Kamm Traffic Characteristics and Communication Patterns in Blogosphere
Overview • Three Perspectives • Server View • All Users, All Blogs • User View • Individual User Perspective • Object View • Individual Blog Perspective
Server View • File transfer exhibits Pareto distribution • Diurnal, bursty patterns • Most blog traffic (~%40) from search engines
User View • Search engines have little impact on blog popularity • Power law relationship of “interest”
Object View • Blog popularity follows power law • Three blog types • Broadcast • Parlor • Register
Data • Almost a terabyte of data spanning a full month • Over 35 million server requests • Extensive data on each request • Eliminate crawlers and errors • Even takes administrative requests into account
All Views Analyzed • Takes all perspectives into account • Useful for infrastructure side, user experience and social networking • Broad scope
New Interesting Findings • Search engine have little impact on object popularity • Author/reader relationship categorization • Blogosphere patterns more dependent on social networks than traditional web traffic
Relevance / Applications • Synthetic traffic generation • Track blog popularity using owner's social attributes not other pages pointing to it