Web Graph
Giant Global Graph
图论、线性代数若干概念回顾
本次课大纲
Web的大小 —网页总数
Capture-Recapture Model
Web的大小
Overlap analysis
Web的形状
Bow-tie Components
Component Properties
站点入度分布
80/20 原则
Power law
Power law
Power Law Size and Connectivity
Power Law Distribution -Examples
Random Graph. vs. Power Law Graph
Examples with Power Law Networks
What does this mean?
小世界网络
The small world of WWW
Implications for Web
Robustness and vulnerability
对网页重要性的评价
重要度的度量 Paul Erdös
Page. Rank
谁重要一些？
声望模型 Reputation Model
1 2 3
先前那 4个点的例子也无解
即使有解，还有可能不唯一！
Page. Rank
“Random Walker”模型
Random walker model
Random Walker Model (continue)
Stochastic matrix
还有一点问题
Google Matrix特征向量求解
问题
例子（ power iteration）
小规模数据求解
State-of-art Page. Rank Technology
ODP-Biasing
Results
Results
References
本次课小结
第一次编程作业
Thank You!
阅读材料
Steve Lawrence
Larry Page & Sergey Brin
Zipf’s law
Co-citation分析
![coc[i, j]=∑E[k, i]E[k, j]](https://present5.com/presentation/1a161c19499c9d64b3c13e92ab37aec5/image-73.jpg "coc[i, j]=∑E[k, i]E[k, j]")
coc[i, j]=∑E[k, i]E[k, j]
修改模型
解的存在性
The Perron-Frobenius Theorem
Random Walker Model (continue)
HITS
Authority and Hub scores
Power Iteration on:
Overlap analysis
Скачать презентацию Web Graph Giant Global Graph 图论 线性代数若干概念回顾
1a161c19499c9d64b3c13e92ab37aec5.ppt