大学课程《社交大数据分析方法与应用》PPT课件：07-社交网络中的角色、位置和聚类.pptxVIP

社交网络中的角色、位置和聚类;位置、角色和聚类分析;研究目的;A)图论方法（Graphtheoreticalmethods）:派系及其扩展

派系Cliques

K-核k-cores

M-核M-core

K-丛k-plexes

n-clan

K-components

B)程序算法Algorithmicmethods:

Standardmodels:

-Factions(UCI-NET)

-NEGOPY(Richards)

-KliqueFinder(Frank)

-RNM,JIGGLE(Moody)

-BetweenessCentrality(Newman)

-GeneralDistanceClusteringMethods;Extensionsofthisideainclude:;位置与角色;Startwithsomebasicideasofwhataroleis:Anexchangeofsomething(support,ideas,commands,etc)betweenactors.Thus,wemightrepresentafamilyas:;;结构性等价;位置分析;位置分析;;聚类分析;聚类方法;Clusteranalysis;Clusteranalysis;1、定义距离的准则;2、常用距离的算法;(2)绝对值距离或曼哈顿距离

距离为值差的绝对值之和，其计算公式为：

;第二步：计算次远的点集；依次重复。

最后得到嵌套的类型。;Clusteranalysis;datarandom;

doi=1to20;

x=rannor(0);

y=rannor(0);

output;

end;

run;;Resultingdendrogram;;（a）空间的群点(b)任取两个聚核

;(e)第二次分类

;Anexample:

Padgett,J.F.andAnsell,C.K.RobustactionandtheriseoftheMedici,1400-1434.AmericanJournalofSociology.1993;981259-1319.

“PoliticalGroups”intheattributesensedonotseemtoexist,soPAturntothepatternofnetworkrelationsamongfamilies.

Thisistheblockreductionofthefull92familynetwork.;3、相似系数的算法

（1）相似系数

;（2）夹角余弦;PositionAnalysis-Measures;CONCOR算法;JeffersonHighSchool;1.00-0.200.080.08-0.19-0.190.770.770.770.77-0.26-0.26-0.26-0.26

-0.201.00-0.19-0.190.080.08-0.26-0.26-0.26-0.260.770.770.770.77

0.08-0.191.001.00-1.00-1.000.360.360.360.36-0.45-0.45-0.45-0.45

0.08-0.191.001.00-1.00-1.000.360.360.360.36-0.45-0.45-0.45-0.45

-0.190.08-1.00-1.001.001.00-0.45-0.45-0.45-0.450.360.360.360.36

-0.190.08-1.00-1.001.001.00-0.45-0.45-0.45-0.450.360.360.360.36

0.77-0.260.360.36-0.45-0.451.001.001.001.00-0.20-0.20-0.20-0.20

0.77-0.260.360.36-0.45-0.451.001.001.001.00-0.20-0.20-0.20-0.20

0.77-0.26