SAN -- SAGE HANDBOOK 之 开篇Intro

<坤>:元亨。利牝马之贞。君子有攸往，先迷，后得主，利。西南得朋，东北丧朋。安贞吉。

初六：履霜，坚冰至。

六二，直、方、大，不习，无不利。

六三，含章，可贞，或从王事，无成有终。

六四，括囊，无咎无誉。

六五，黄裳，元吉。

上六，龙战于野，其血玄黄。

用六，利永贞。

一个看起来很厉害的SNA scholar的网站：http://www.pfeffer.at/projects.php

有一张很酷的poster，source: http://www.pfeffer.at/projects.php

Chap 2 Intro

What Is a Social Network?

• Social network analysis is neither a theory nor a methodology. Rather, it is a perspective or a paradigm. 它没有predict的功能 只是给我们提供了一种看问题的视角

• A social network is a set of socially relevant nodes connected by one or more relations. node 可以是publication 国家 街区等等

• Boundary的选取就很重要了，也就是选取这个网络里member都是谁. 大概三个方法 [Laumann et al. (1983)]：1） position-based. 也就是 在那个网络中有一个position/member；
  2） event-based. who had participated in key events 比如谁参加了指定的那几次会议；3）relation-based. 比如一篇文章的引与被引

• 接着就是这些nodes之间关系的识别。Borgatti et al. (2009) identify four broad categories of relations: similarities, social relations(role like kinship or feelings like liking and knowing), interactions (like helping inviting) and flows (exchanges or transfers).

Guiding Principles of Network Analysis

Relations, Not Attributes

意思就是 在解释一件事情时，经常会用到事物的属性差异来解释。比如individuals中可能是你的种族 受教育程度 等等造成了某种选择的偏差。

在SNA中，在解释causation中更加强调的是social structure而不是attributes。比如 具有相似属性的这些人可能有着相似的 network positions, 这些position的constraint 机会 perception 等等 带来了某些相似的outcome

这样SNA的角度可能就更能说明多actor中某种属性形成的原因 他们之间的关系

Networks, not groups

也就是说不要把nodes当成sets of mutually exclusive groups 以及过度简化
要思考individuals 在 groups 里 的 varying degrees 以及从而产生的不同的responses

Relations in a Relational Context

分析某对关系时 可能也需要考虑到ta们和其他 的关系 比如 要了解罗密欧与朱丽叶的关系 也需要了解他们各自家族的关系

The Origins and Current State of Social Network Analysis

Simmelian Roots

追溯到 work of Georg Simmel, 清楚得表明 social ties are primary. " society itself is nothing more than a web of relations"
He argues we should focus instead on the emergent consequences of the interactions of individual actions

他把这些interaction的pattern称为form，individual motives emotions thoughts feelings beliefs 等称为content.

he argues that: Only by studying similar forms across diverse contents can people truly understand how these forms function as forms and separate the effects of forms from the effects of contents.

Current State: Association, Grants and Journals

The International Network for Social Net work Analysts (INSNA), founded by Barry Wellman in 1977
INSNA's annual conference, the International Sunbelt Social Network Conference
Social Science Research Council of Canada

American Journal of Sociology, American Sociological Review, Social Forces, Human Organization and Administrative Science Quarterly, as well as specialised journals, such as City and Community, Work and Occupations and Information, Communication and Society. Three peer-reviewed journals publish social network research exclusively: Social Networks (INSNA's flagship journal), Connections (an INSNA journal publishing short, timely papers) and the Journal of Social Structure, published online.

Applying a Network Perspective

Formalist Theories

• primarily with pure form – in the mathematical, platonic sense – of networks 主要是数学操作
• they can be studied without the need for empirical data.
• 出名的有 barabsi和Albert的 Matthew effect，watt的six degrees，barabasi的linked，buchanan的nexus

Structuralist Theories

• concerned with how patterns of relations can shed light on substantive topics within their disciplines.
• at least four different approaches：1）Defining Key Concepts in Network Terms 即是 用network概念来给一些关键的概念赋予新的含义，比如 community 可能原来是从地理层面了解，现在更多是看怎么连接；2）Testing an Existing Theory. 3）Looking at Network Causes of Phenomena of Interest；4）Looking at Network Effects of Phenomena of Interest

Network explanations

We follow Borgatti et al.'s (2009) classification of network arguments into four categories: transmission, adaptation, binding and exclusion (see Borgatti, this volume).

Transmission

也就是 spread flow diffusion

Adaptation

即 当两个人做出同样的选择是因为有相似的network positions 从而 exposed 在相似的constraints 机会里。（这里我怎么觉得像是教人怎么透过现象看本质的感觉 never mind)

Binding

即 一个network内部 bind together 并且 act like one

Exclusion

即 排斥 当一个firm 有两个潜在的买家时 可以游刃有余 且增加了 bargaining power

Studying and Operationalizing Networks

key premise of network analysis – that relations are primary

Collecting Network Data

• first decide what kinds of networks and what kinds of relations they will study
• two important dimensions along which network data vary: whole versus ego networks, and one-mode versus two-mode networks.

whole versus ego networks

• whole: 更加关注所有的nodes，而不是有优先的nodes。两个很著名的例子：
  1）一个工厂里所有工人 康康谁和谁玩
  

2. 就是那个movie/star network, Watts, 1999

• egocentric: 关注 ego. Data are on nodes that share the chosen relation(s) with the ego and on relations between those nodes.

One-Mode Data Versus Two-Mode Data

当涉及到group memberships的时候 就需要 two mode 或者affiliation了（这个也强调了很多次）

Types of ties

有向无向 binary or valued

Survey and Interview Methods

data搜集可以通过 观察 archive 历史材料 电子文件的trace等等

Ego network data are most commonly collected using name generators: survey questions that ask respondents to list the people with whom they share a particular relation (Marsden, 1987, 2005; Burt, 1984; Hogan et al., 2007).

survey 或者 interview的过程是很漫长的，而设计 也是 比如问卷需要很复杂的 skips and loops等等

Analysing Network Data

Chap 3 The Development of Social Network Analysis – with an Emphasis on Recent Events

首先回顾一下作者之前总结的 SNA 四个重要的properties （Freeman 2004）：

(1) It involves the intuition that links among social actors are important; 连接很重要

(2) it is based on the collection and analysis of data that record social relations that link actors; 数据很重要

(3) it draws heavily on graphic imagery to reveal and display the patterning of those links; 可视化很重要

(4) it develops mathematical and computational models to describe and explain those patterns. 数学CS很重要

（55 然而我一个技能都不会）

The modern field of social network analysis, then, emerged in the 1930s，这个时段大概有三个很厉害的team：

随后到七十年代的这段期间：

不过 仍旧 没有哪个center可以提供一个generally recognized的paradigm ok 然后 early 1970s 情况变了！

Harrison C. White

Following the contributions of White and his students, social network analysis settled down, embraced a standard paradigm, and became widely recognized as a field of research. （这个评价有点牛逼）

时间来到了1990s，revolutionary change来了：物理学家 开始 publish on social networks

(然后作者开始批评这几个物理学家 根本就没有仔细地读他们之前的东西，一点也不像其他的物理学家blablabla 但是承认了这两篇文章所带来的巨大的影响—）

The Origins of the Revolution

Small world by Watts and Strogatz

这个model 首先是试图给一堆朋友们capture 聚类，然后定义一个 average clustering coefficient C(p)，如下左图：

结果是 C(p) 很高，同时 average length of the path L(p) 也很高，左图就不是small world。于是他们去掉了close neighbors 之间的连线并加入了随机的连线，从而到了右图，which L(P) 减小很多 但 C(P)几乎没变，也就是说 图中的朋友们还是朋友，但整个世界变小了：

Distribution of connections by Barabási and Albert

他们给出的例子里包含WWW actors electric power grid等等 然后发现 networks中的links是skewed！并且 这些connection的分布符合幂定律 也就是 "scale free."

The Growth of the Revolution

直到 2004年 （作者研究的节点）物理学家发表了非常多的小世界论文 而不是social network community,同样 社会网络分析家也是。 Gradually，物理学家们开始拓展到其他各种内容， 生物学家 cs也参与进来。

Cohesive Groups or Communities

The notion of cohesive groups is foundational in sociology.

structural perspective on groups [Freeman and Webster (1994)] :

Whenever human association is examined, we see what can be described as thick spots—relatively unchanging clusters or collections of individuals who are linked by frequent interaction and often by sentimental ties. These are surrounded by thin areas where interaction does occur, but tends to be less frequent and to involve very little if any sentiment.

Models of cohesive groups: algebraic, graph theoretic, matrix permutation.

Over the years social network analysts have also drawn on various computational algorithms in an attempt to uncover groups. These include multidimensional scaling (Freeman et al., 1987; Arabie and Carroll, 1989); various versions of singular value decomposition, including principal components analysis and correspondence analysis (Levine, 1972; Roberts, 2000); hierarchical clustering (Breiger et al., 1975; Wasserman and Faust, 1994: 382–83); the max-cut min-flow algorithm (Zachary, 1977, Blythe, 2006); simulated annealing (Boyd, 1991: 223; Dekker, 2001); and the genetic algorithm (Freeman, 1993; Borgatti and Everett, 1997). 降维 核 主成分分析 聚类 优化问题 搜索策略问题

总体而言，SNA 逐渐转向建立在概率的问题上 （我逃不过数学 哭了）

关于 最近物理学家的兴趣可以看 Newman and Girvan 研究 edge betweenness degree partition modularity。开始这个 betweeness-based 算法 computationally slow，后来他们又研究出了一种 fast greedy algorithm

The concern with computing speed seems to have started a race to see who could develop the fastest algorithm to cluster nodes in terms of their modularity. 对于实际应用中 速度非常关键

Positions

大概有四种positions：

1）positions in groups — core and periphery — have been specified. 是核心呢还是边缘呢

• in a pair of articles, (Borgatti and Everett, 1999; Everett and Borgatti, 2000) two researchers developed a full model of core/periphery structure.

2）a good deal of attention has been focused on social roles. 实际的社会角色是什么呢

• introduced by the anthropologist Ralph Linton (1936) 父亲和孩子

3）some attention has also been devoted to the study of the positions of nodes in hierarchical structures. 层级结构中又是什么角色呢（这个在policy看到好多）

• hierarchies or dominance orders，began with Pierre Huber's (1802) observations of dominance among bumblebees.

• Martin Landau (1951), who was both an ethologist and a social network analyst, created a formal model of hierarchical structure for social network analysts.

• James S. Coleman (1964), proposed an alternative model

4）social network analysts have been concerned with the structural centrality of nodes in network 就是对于各种中心度的思量