《Modeling Blog Dynamics》.pdfVIP

Modeling Blog Dynamics ¨ ∗ ∗ Christos Faloutsos Michaela Gotz Jure Leskovec Mary McGlohon Cornell University Cornell/Stanford University Carnegie Mellon University Carnegie Mellon University goetz@ jure@ mmcgloho@ christos@ Abstract static web pages (Dezs¨o et al. 2006; Ntoulas, Cho, and Ol- ston 2004). Earlier work has found surprising patterns in How do blogs produce posts? What local, underlying mech- blog dynamics: there are unexpected power laws in the pop- anisms lead to the bursty temporal behaviors observed in ularity of blogs and the distribution of blog sizes, and self- blog networks? Earlier work analyzed network patterns of similar (and bursty) patterns in the blog activity. Our goal is blogs and found that blog behavior is bursty and often fol- lows power laws in both topological and temporal character- to understand the mechanisms in individual blogs that gen- istics. However, no intuitive and realistic model has yet been erate these patterns. introduced, that can lead to such patterns. Applications of our work include modeling blog popular- This is exactly the focus of this work. We propose a gener- ity and information diffusion in the blogosphere. The re- ative model that uses simple and intuitive principles for each sults of the mod