Random Graph Models of works：社交网络讲义的随机图模型
医学网络平台分析 随机游走
Random Walk是扩散过程的基础，因此它被广 泛地用于对物理和化学等扩散现象的模拟上。
在应用数学上，Random Walk是利用随机因 素模拟真实的动力系统的常用方法。图 (a)和 (b)分别显示的是Random Walk在二维和三维 空间的模拟图。
此外，Random Walk又是设计随机算法的一 个非常广泛的工具，其中一个典型的例子就是 “马尔可夫链蒙特卡罗”法（MCMC）。 MCMC是解决近似计算问题一种重要方法，它能 以比确定性算法快指数级的速度提供解决问题的 最好随机方法，目前已经被广泛地应用在统计领 域。
W:图的列向量归一化的邻接矩阵。 P0:初始概率向量，对于每一个种子结点，以相
等的概率分配，向量中各元素之和为1。
r:游走者返回种子结点的概率（重启动概率）， r的值越接近于0，表示随机游走方法考虑图的
全局结构特征越多。
P t 1
0
t
t
t
t
5 0.1 * P5 0.9 * P8 * (1 / 4) 0.9 * P6 * (1 / 2) 0.9 * P7 * (1 / 2) 0.9 * P4 * (1 / 3)
随后，Brin 和Page 利用PageRank 技术成 立了Google 公司，而今天Google已经在人们 的日常生活中显示出巨大影响，所以可以毫不 夸张地说Random Walk改善了今天的生活。下 图为PageRank 的一般描述示意图，边上显示 的是转移概率，而网页上除了ID 外显示的是页 片的Rank 值。
这里我们给出一些可以参考的理论著作：一 个比较好的关于图上的Random Walk 的理论 研究综述是由Lovász 给出的;
Lovász, L, Random walks on graphs: a survey. In Combinatorics, Paul Erdos is eighty, Vol.2, pp. 353–397, 1993.
《社交网络分析》PPT课件
Degrees of Separation）
最多通过六个人，你可以认识任何陌生人
社交拓展的形式
根据熟人的熟人聚集：瓣 根据学习经历进行聚集：Facebook、人人网
1、确认案例分析对象
18
完整版ppt
2、案例分析问题提出
19
完整版ppt
能否找到一种方法自动地为我的所有好友进行分组？？
20
完整版ppt
3、案例解决方案思路 社交网络是一个复杂的人际关系网络 物以类聚，人以群分 采集社交网络人际关系数据，进行聚类分析，发现
群组
21
工欲善其事，必先利其器——开发语言选择
完整版ppt
Python，是一种解释型的，面向对象的、带有动态语义 的高级程序设计语言。（）
Python设计哲学：优雅、明确、简单。
最好是只有一种方法做一件事
Python可以做什么：
系统编程
用户图形接口
Internet脚本
组件集成
数据库编程
快速原型
数值计算与科学计算编程
16
完整版ppt
三、社交网络分析案例介绍
1、确认案例分析对象 2、案例分析问题提出 3、案例解决方案思路
工欲善其事，必先利其器——开发语言选择
复杂网络处理程序库 4、案例解决方案实现
社交网络数据采集方法介绍 简单网络爬虫过程分析 社交网络数据采集结果展示 社交网络社群发现
17
完整版ppt
帮助企业了解：
哪些客户或群体中的哪些成员能够影响其他客户和事情 的前景
哪些人是潜在的，将严重影响客户满意度的人 哪些人在推动产品和技术的革新方面拥有特殊的影响力
《大数据分析》课件-第13章 社交网络与推荐系统
C
图中有5个实体及其间的4段关系
13.2.1
社交网络的 统计学构成
13.2.2
社交网络的 群体形成
13.2.3 图与网络分析
13.2 社交网络的结构
网络是可以描述自然和社会的大 规模的系统，这些系统包含的信 息丰富多样，结构也更加复杂， 通常建模后会形成复杂网络。
13.2.1 社交网络的统计学构成
13.2.1 社交网络的统计学构成
一些统计学中社交网络的相关研究和理论，例如： （1）随机图理论。随机图的“随机”体现在边的分布上。一个随机图是将给定的顶点之 间随机地连上边。假设将一些纽扣散落在地上，并且不断随机地将两个纽扣之间系上一 条线，这就得到一个随机图的例子。边的产生可以依赖于不同的随机方式，产生了不同 的随机图模型。
在网络理论的研究中，复杂网络是由数量巨大的节点和节点之间错综复杂的关系共同构 成的网络结构，用数学语言来说，就是一个有着足够复杂的拓扑结构特征的图。复杂网 络分为随机网络、小世界网络和自相似网络。小世界网络和自相似网络介于规则和随机 网络之间。 复杂网络具有简单网络（如晶格网络、随机图）等结构所不具备的特性，而这些特性往 往出现在真实世界的网络结构中。复杂网络的研究是现今科学研究中的一个热点，与现 实中各类高复杂性系统（如互联网、神经网络和社交网络）的研究有密切关系。
大数据与人工智能有着千丝万缕的关系，互联网公司一般会构建自己的大数据与人工智 能团队，构建大数据基础平台，基于大数据平台构建上层业务，包括商业智能（BI）， 推荐系统及其他人工智能业务，右图是典型 的基于开源技术的视频互联网公司大数据与 人工智能业务及相关的底层大数据支撑技术。
大数据支撑下的人工智能技术体系 （DS：数据源，DC：大数据中心， BIZ：上层业务）
sbm模型 松弛度量方法 -回复
sbm模型松弛度量方法-回复sbm模型是社交网络分析中常用的一种模型，用于将网络中的节点分成若干个社区。
而松弛度量方法则是一种通过衡量社区结构之间的相似度来评估分区质量的方法。
本文将会详细介绍sbm模型的原理和应用，并以松弛度量方法作为评估指标，一步一步回答该主题。
第一部分：sbm模型的原理和应用1.1 sbm模型的原理sbm模型全称为Stochastic Block Model，是一种随机图模型，在社交网络中被广泛应用于社区发现。
它基于一个假设，即网络中的节点可以被划分为若干个社区，并且同一社区的节点具有相似的连接概率，不同社区之间的连接概率较低。
1.2 sbm模型的应用sbm模型在社交网络分析中有许多应用，例如：- 社区发现：通过将网络中的节点划分为不同的社区，sbm模型可以帮助我们了解网络中的群组结构和关联关系，为后续的行为预测、目标推荐等任务提供基础。
- 信息传播：sbm模型可以帮助我们预测信息在网络中的传播路径和速度，进而优化信息推送策略和社交广告投放。
- 社交影响力分析：通过sbm模型，我们可以分析社交网络中不同节点的影响力大小，了解网络中的意见领袖和关键节点，从而更好地设计营销策略和舆情监测。
第二部分：松弛度量方法的介绍2.1 松弛度量方法的定义松弛度量方法是一种通过计算不同社区结构之间的相似度来评估分区质量的方法。
它基于一种假设，即对于一个理想的社区划分，同一社区的节点应该更有可能彼此相连，不同社区的节点联系较少。
2.2 松弛度量方法的计算松弛度量方法通常基于社区结构之间的连接概率计算，常用的指标包括：- 模块度（Modularity）：衡量实际社区划分与随机模型之间的差异程度，数值越大表示社区结构越好。
- 规模比（Size Ratio）：衡量实际社区划分中不同社区规模大小的平衡程度。
- 新颖性（Novelty）：衡量实际社区划分中新出现的社区数量，用于评估算法的发现能力。
第三部分：sbm模型与松弛度量方法的应用示例为了更好地理解sbm模型和松弛度量方法的应用，我们以一个社交网络为例进行说明。
【会议文章】时空社交网络的计算模型：一种时间聚合的图形法
Computational Modeling of Spatio-temporal SocialNetworks: A Time-Aggregated Graph ApproachS HASHI S HEKHAR AND D EV O LIVERDepartment of Computer Science & EngineeringUniversity of Minnesota, MinneapolisEmail: shekhar@; Web: /~shekharIntroductionSocial computing is transforming on-line spaces with popular applications such as social-networking (e.g., Facebook), collaborative authoring (e.g., Wikipedia), social bargain-hunting (e.g., Groupon), etc. Spatio-temporal constraints are becoming a critical issue in social computing with the emergence of location-based social-networking, Volunteered Geographic Information (Goo 07, Elw 08), Participative Planning (Elw 08, Fis 01), etc. Location-based social networks (e.g., and the “Places Check-in” feature on Facebook) facilitate socialization with nearby friends at restaurants, bars, museums, and concerts. Volunteered Geographic Information (e.g., Wikimapia, OpenStreetMap, Google MyMaps) allows Internet users to participate in generation of geographic information. Traditional computational models for social networks are based on graphs [Fre 06, Was 94, Nrc 03, Cro 09], where nodes represent individual actors (e.g., persons, organizations) and edges represent relationship ties (e.g., communication, financial aid, contracts) between actors. Such graph models are used to assess centrality and the influence of actors (e.g., measures such as degree, reach, “between-ness,” bridge), as well as community structure (e.g., measures such as cohesion, clustering, etc.). Statistical properties such as skewed degree distribution are modeled by random graphs [New 02, Nrc 03], where each node-pair has a connecting edge with independent probability p, which may depend on factors such as geographic distance [Won 05].However, traditional graph and random graph models are limited in addressing spatio-temporal questions such as change (e.g., how is trust or leadership changing over time? who are the emerging leaders in a group? what are the recurring changes in a group?), trends (e.g., what are the long-term and short-term trends in network size or structure? what are the exceptions to the long-term trend?), duration (e.g., how long is the tenure of a leader in a group? how long does it take to elevate the level of trust such as a relationship changing from visitor to friend?), migration, mobility and travel (e.g., interplay between travel behavior and size/structure of social networks [Tim 06]). This position paper explores time-aggregated graph models to support computational tools to address such questions.Background“A social network is a social structure made up of individuals (or organizations) called “nodes,” which are tied (connected) by one or more specific types of interdependency, such as friendship, kinship, common interest, financial exchange, dislike, sexual relationships, or relationships of beliefs, knowledge or prestige” [Wik 10]. Network formation theories bringup the principle of homophily [McP 01] (i.e. birds of a feather flock together) and differentiate between two types, namely, baseline and in-breeding. Baseline homophily [Fel 81] refers to the limited social pool available to individuals for tie formation due to foci of activities, demographics, etc. In-breeding homophily models additional constraints due to gender, religion, social class, education, personality, etc.Spatio-temporal constraints (e.g., geographic space, travel, schedules and diurnal cycles) play a major role in determining baseline homophily due to reasons like opportunity and minimization of cost and effort [Deb 69]. Survey research on student housing communities [Ath 73, Mok 04] and Torontonian personal communities [Wel 88] have provided evidence on the role of geographic spaces. For example, [Wel 88] noted that about 42% of “frequent contact” ties lived within a mile of a typical person. Computational simulation based on agent-based models [Cro 09], game-theory and cost-benefit analysis [Joh 00] as well as spatially embedded random graphs and distance-decay based edge probabilities [Won 05] have reproduced many properties of social networks including small-world properties (e.g., graph diameter), short average geographic distances, community structures, low tie density, etc.Traditionally, social network research has had a relatively small amount of data from infrequent longitudinal surveys and computer simulations. However, the recent popularity of social computing on the Internet is providing large and frequently-sampled spatio-temporal social interaction datasets. These may facilitate better understanding of social network formation and the role that spatio-temporal constraints play, in the face of opportunities to interact with distant actors via Internet based social networking applications. This development also highlights a central role for computation and computational models, not only to scale up to the large and growing data volumes, but also to address new spatio-temporal social questions related to change, trends, duration, mobility, and travel.Time-Aggregated GraphsGiven a spatio-temporal social network dataset and analysis questions, a computational model provides a representation to not only specify the dataset and questions, but also design data-structures and algorithms for addressing the questions. The model should be able to scale as large datasets are becoming available from Internet based social computing services with a large number of actors and a large number of time-points.There are several challenges associated with modeling such a network. The model should be able to accommodate changes and compute results consistent with the existing conditions both accurately and simply. Furthermore, for quickly answering frequent queries such as tracking recurring changes in a group, fast algorithms are required for computing the query results. Sufficient support for the design of correct and efficient algorithms should therefore be provided by the model. The need for computational efficiency conflicts with the requirement for expressive power of the model and balancing these two conflicting goals is challenging.Figure 1: A Time-series of Snapshots for a trust Network for time instants 1 through 10Dynamic networks are often modeled using a time-series of snapshots [Tan 07, Lah 08] of actors and their relationships. For example, Figure 1 shows a time-series of snapshots of a trust network for time instants, t = 1 to t = 10. Three levels of trust are exhibited in Figure 1, where an absent edge indicates that there is no trust relationship, an edge with V indicates a visitor trust relationship and an edge with F indicates friendship, which is a stronger relationship than visitor. For example, (N1, N3) have no trust relationship at t = 1, 2, and 4, a visitor relationship at t = 3, 5, 6, and 7 and a friend relationship at t = 8, 9, and 10. Due to duplication of information about nodes and edges across snapshots, the scalability of snapshot time-series model is limited in answering longitudinal questions such as how long it takes a relationship to evolve from visitor to friend. Storage and thus computational cost increases linearly with the number of time-points.An alternative is the time aggregated graph (TAG) model [Geo 06], which has provided scalable algorithms for temporal questions in transportation networks [Geo 08]. Figure 2 shows a time aggregated graph that is based on the trust network of Figure 1. Each edge has a time series (enclosed in square brackets). For example, the trust relationship for the edge (N1, N3) for all instants within the time interval under consideration are aggregated into a time series [-,-,V,-,V,V,V,F,F,F]; the entry ’-’ indicates that the edge is absent at the time instants t = 1, t = 2 and t = 4.Figure 2: A Time Aggregated Graph for a Trust Network for time instants 1 through 10With the time aggregated graph model, the longitudinal behavior is captured for each edge (or node). Time aggregated graphs do not replicate information about nodes and edges across time-points. This reduces storage cost as well as computational costs. It also consolidates time-series information to make it easier to answer cross snapshot questions such as how long it takes for a relationship to evolve from visitor to friend. For example, in Figure 3, it takes 2 units for edge (N1, N2), 3 units for edge (N2, N4), 4 units for edge (N1,N3) and 1 unit for edge (N3, N4) with an average of (2 + 3 + 4 + 1) / 4 = 2 units, for a relationship to evolve from visitor to friend. Answering this duration question in snapshot model takes more effort due to the need to repeatedly go from one snapshot to the other.DiscussionTime-Aggregated Graphs has the potential to be a general representation of temporal evolution of relationships, whose snapshots were traditionally modeled as graphs. Thus, it may help address questions about individual relationships over longer time-frames by providing a representation for relevant datasets. For example, consider social network datasets representing friend-of relationships among people. Currently, graph representations are used to model a static (e.g., time-snapshot) view of social relationships to explore questions like centrality (e.g., leaders), and cohesiveness of communities. In contrast, TAG may support a direct representation of a “friend-of” relationship over long time periods to address questions related to changes in and evolution of centrality (e.g., emerging leaders) and group cohesiveness (e.g., increasing, diminishing). We welcome collaboration towards identifying datasets and use-cases to evaluate the potential of TAG to address spatio-temporal questions about social networks.References:[And 08] C. F. d’Andrea, Wikipedia as social interaction rooms and collaborative-writing processes, WikiSym, ACM 978-1-60558-128, 2008.[Ath 73] R. Athanasiou, G. A. Yoshioka, The spatial characteristics of friendship formation, Environ.Behav, 5, 1973, 43–66.[Cro 09] Crossman, J., Bechtel, R., Parunak, H., and Brueckner, S., Integrating dynamic social networks and spatio-temporal models for risk assessment, wargaming and planning. In The Network Science Workshop. 2009, West Point, NY.[Deb 69] G. Debreu, Neighboring economic agents, La Decision, 171: 85–90, 1969.[Elw 08] S. Elwood, Volunteered Geographic Information: Future Research Directions Motivated by Critical, Participatory, and Feminist GIS. GeoJournal 72(3 & 4): 173–183, 2008.[Fel 81] S. L. Feld, The focused organization of social ties, Am. J. Soc., 86, 1981, 1015–1035.[Fis 01] F. Fisher, Building Bridges through Participatory Planning, UN-Habitat, isbn 92-1-131623-5, 2001. /decision/BuildingBridges.pdf.[Fre 06] L. Freeman, The Development of Social Network Analysis. Vancouver: Empirical Pres, 2006. [Geo 08] B. George, S. Shekhar, and S. Kim. Spatio-temporal network databases and routing algorithms. University of Minnesota. CSE Technical Report No. 08-039, 2008. (A summary of results in Proc. Symposium on Spatial and Temporal Databases, Springer LNCS 4605, 2007).[Geo 06] B. George, S. Shekhar, Time-Aggregated Graphs for Modeling Spatio-temporal Networks, Journal of Data Semantics, 11, 2006 (Springer LNCS 5383, 2008). (Special issue on best papers from ER 2006 Workshop on Conceptual Modeling for GIS).[Goo 07] M. F. Goodchild, Citizens as sensors: The world of volunteered geography. GeoJournal, 69(4): 211–221, 2007.[Joh 00] C. Johnson and R. P. Giles, Spatial Social Networks, Review of Economic Design, 5, 273–299, Springer Verlag, 2000.[Lah 08] M. Lahiri, T. Y. Berger-Wolf, Mining Periodic Behavior in Dynamic Social Networks, Proc.ICDM, IEEE, 2008.[McP 01] M. McPherson, L. Smith-Lovin, J. M. Cook, Birds of a feather: Homophily in social networks, Annu. Rev. Sociol., 27, 2001, 415–444.[Mok 04] D. Mok, B. Wellman, R. Basu, Does distance matter for relationships? SUNBELT International Social Network Conference, Portoroz, Slovenia, 2004.[New 02] M. E. Newman, D. J. Watts, S. Strogatz, Random graph models of social networks, Proceedings of the National Academy of Science, 99, 2566–2572, Feb. 19th, 2002.[Nrc 03] National Research Council, Dynamic Social Network Modeling and Analysis: Workshop Summary and Papers, isbn 0-309-08952-2, National Academies Press, 2003.[Tan 07] C. Tantipathananandh, T. Berger-Wolf, D. Kempe, A framework for community identification in dynamic social networks, Proc. SIGKDD, ACM, 2007.[Tap 06] D. Tapscott, A. Willians. Wikiconomics: How mass collaboration changes everything. New York: Portifolio, 2006.[Tim 06] O. Timo, Mapping Social Networks in Time and Space, Arbeitsberichte Verkehr und Raumplanung, Working Paper 341, IVT, ETH Zurich, 2006.[Was 94] S. Wasserman, and K. Faust, Social Network Analysis: Methods and Applications, Cambridge University Press, Cambridge, 1994.[Wel 88] B. Wellman, P. Carrington, A. Hall, Networks as personal communities, in B. Wellman, S. D.Berkowitz (Eds.), Social Structures: A Network Approach, Cambridge University Press, Cambridge, 1988.[Wik 10] Social Network, Wikipedia, /wiki/Social_network, retrieved Nov. 29, 2010.[Won 06] L. H. Wong, P. Pattison, G. Robbins, A spatial model for social networks, Physica A, 360, 2006, 99–120, Elsevier.。
nx.random_geometric_graph的用法
nx.random_geometric_graph的用法一、概述nx.random_geometric_graph 是 NetworkX 的一种生成随机图的函数，主要基于平面上点的几何距离来连接节点，也称为基于几何距离的随机图生成器。
其生成的随机图常被用于测试网络算法和进行实验。
二、函数定义```python nx.random_geometric_graph(n, radius, dim=2, pos=None, p=2, periodic=False, seed=None)```参数：- n ：图中节点的数量 - radius ：表示点之间的最大距离。
- dim ：表示空间的大小，例如dim=2表示空间为二维平面。
- pos ：表示节点的位置，默认随机生成。
- p ：好像是一个距离指数，决定边的形成情况，当p → ∞ 时，该随机图趋近于完全图，反之则趋近于星形图。
- periodic ：表示是否使用周期性边界。
- seed ：为随机数生成器设置一个种子。
返回值：生成一个随机图。
三、基本用法1.用法一：简单使用只需要指定节点的数量、半径以及维数即可得到这个随机图。
```python import networkx as nxn = 50 radius = 0.2 G =nx.random_geometric_graph(n, radius) ```2.用法二：搭配pos参数使用pos 表示节点的位置，可以指定节点在二维平面上分布的位置。
可以利用 numpy 库生成一些二维坐标作为参数传递，这种情况下我们不需要再指定维数和半径。
下面是一个例子：```python import numpy as np import matplotlib.pyplot as pltpos = np.random.rand(n, 2) # 在二维平面中随机生成n个坐标 G = nx.random_geometric_graph(n,pos=pos) nx.draw(G, pos, node_size=20,node_color='blue') plt.show() ```我们可以发现，节点在正方形的边界内随机分布，没有直接相连的节点。
生物分子网络 第二章
第二章一.名词解释1.全局耦合网络：任意两个点之间都有边直接相连，完全连接。
2.最近邻耦合网络：每一个节点只和它周围的邻居节点相连。
3.星形耦合网络：有一个中心点，其余N-1个点都只与这个中心点连接4.均匀网络：当k >> <k>时，度为k的节点几乎不存在。
因此这类网络也成为均匀网络或指数网络5.无标(尺)度网络：由于这类网络的节点连接度没有明显的特征长度6.随机网络：节点度的分布将遵循钟形曲线分布。
按照这种分布，大多数节点拥有的连接的数目都相差不多7.鲁棒性：如果移走少量节点后，网络中的绝大部分节点仍是连通的，那么称该网络的连通性对节点故障具有鲁棒性或者稳健性。
8.脆弱性：蓄意去除少量度最高的节点就可破坏无标度网络的连通性9.设计网络：随机网络中节点总数N是预先给定的，所以它们是静态的、固定的、平衡的网络，也有称为设计网络10.演化网络：若网络模型的节点总数不是预先给定的，而是逐步增减的，则它们是动态的、增长的、非平衡的网络，或者称为演化网络(evolving network)11.马太效应：新的节点更倾向于与那些具有较高连接度的“大”节点相连接，这种现象也称为“富者更富（rich get richer）”或“马太效应(Matthew effect)”。
12.分形几何：普通几何研究的对象一般都具有整数的维数，比如，零维的点、一维的线、二维的面、三维的立体、乃至四维的时空。
分形几何（fractal geometry）是研究具有不一定是整数的维，而存在一个分数维数的空间。
13.适应度：在许多实际网络中，节点的度及其增长速度并非只与该节点的年龄有关，有时是与节点的内在性质有关的，Bianconi和Barabasi把这一性质称为节点的适应度（fitness）14.模块：模块（model）是指一组物理上或功能上连接在一起的、共同完成一个相对独立功能的节点。
15.模体：具有高聚类性的网络在局部可能包含各种由高度连接的节点组构成的子图（subgraph），如三角形，正方形和五角形，其中一些子图所占的比例明显高于同一网络的完全随机化形式中这些子图所占的比例，这些子图就称为模体。
一文读懂社会网络分析(SNA)理论、指标与应用
一文读懂社会网络分析(SNA)理论、指标与应用开新坑！社交网络分析（又称复杂网络、社会网络，Social Network Analysis）是诞生于数学图论、计算机科学、物理学的交叉碰撞中的一门有趣的学科。
缘起：我研究SNA已经有近2年的时光，一路坎坷走来有很多收获、踩过一些坑，也在线上给很多学生讲过SNA的入门知识，最近感觉有必要将心得和基础框架分享出来，抛砖引玉，让各位对SNA感兴趣的同学们一起学习进步。
我的能力有限，如果有不足之处大家一起交流，由于我的专业的影响，本文的SNA知识可能会带有情报学色彩。
面向人群：优先人文社科类的无代码学习，Python、R的SNA 包好用是好用，但是对我们这这些社科的同学来说门槛太高，枯燥的代码首先就会让我们丧失学习兴趣。
特征：类综述文章，主要目的是以通俗的语言和精炼的框架带领各位快速对SNA领域建立起一个全面的认知，每个个关键概念会附上链接供感兴趣的同学深入学习。
开胃菜：SNA经典著作分享《网络科学引论》纽曼 (访问密码 : v9d9g3)2 概述篇：什么是网络？我们从哪些角度研究它？1) 认识网络SNA中所说的网络是由节点（node，图论中称顶点vertex）和边（edge）构成，如下图。
每个节点代表一个实体，可以是人、动物、关键词、神经元；连接各节点的边代表一个关系，如朋友关系、敌对关系、合作关系、互斥关系等。
最小的网络是由两个节点与一条边构成的二元组。
Les Miserables人际关系网络2) 构建网络就是建模马克思说过，“人的本质在其现实性上，它是一切社会关系的总和。
” 事实上，当我们想快速了解一个领域，无论该领域是由人、知识、神经元乃至其他实体集合构成，利用SNA的方法将实体及其相互关系进行抽象和网络构建，我们就完成了对某一领域的“建模”，这个模型就是网络图，拿科学网络计量学家陈超美的观点来说，借助网络图，“一图胜千言，一览无余”。
3) 社会网络类型此处展示常见且常用的网络类型名词，想要具体了解可以点击链接仔细查看！•网络中节点的来源集合异同o一模网络 one-modeo二模网络 two-mode•视角：•边权重o加权网络 weight networko无权网络 unweight networko符号网络 Signed network•关系是否有方向o有向网络 Directed networko无向网络 Undirected network4) 网络分析的5大中心问题SNA可以帮助我们快速了解该网络中的分布格局和竞争态势，“孰强孰弱，孰亲孰远，孰新孰老，孰胜孰衰”，这16字箴言是我学习SNA总结的精华所在，初中级甚至高级的社会网络分析学习几乎完全就是围绕着这四个方面开展，后面将要讲到的理论与方法皆为此服务，希望同学们可以重点关注。