网络论坛的自相似性及其模型

网络论坛是一种主要的互联网应用，人们对它的研究主要集中在话题分析、变化趋势分析等方面。该文研究网络论坛中文章数随
时间变化的统计特性，通过方差分析、R/S 分析方法发现自相似性存在于网络论坛中。为描述这种自相似性，提出一种基于时延厚尾分布
的产生式模型，对该模型进行理论与仿真分析，验证了该模型具有表达网络论坛自相似性的能力，模型的计算复杂度小。
关键词：网络论坛；自相似性；产生式模型
Self-similarity and Modeling of Web-based Forum
ZENG Jian-ping, ZHANG Shi-yong
(School of Computer, Fudan University, Shanghai 200433)
【Abstract】Web-based forum is one of important applications on Internet. More and more research has been focused on it, such as topic analysis,
trend analysis. This paper concentrates on the statistical character of the posting number in Web-based forum. By means of variance analysis and R/S
analysis, finds that self-similarity generally exists in Web-based forum. A generative model based on time-delay levy distribution is proposed to
describe the kind of property. Theoretic and simulation analysis are done to verify the effectiveness of the model and show that the computation
complexity is low.
【Key words】Web-based forum; self-similarity; generative model