荣老师倾情推荐,Watts 关于实验室进行虚拟博弈实验的一个报告。 Speaker Duncan Watts Host Jennifer Chayes Affiliation Microsoft Research New York City Duration 01:15:22 Date recorded 5 December 2012 Crowdsourcing sites like Amazon's Mechanical Turk are increasingly being used by researchers to construct virtual labs in which they can conduct behavioral experiments. In this talk, I describe some recent experiments that showcase the advantages of virtual over traditional physical labs, as well as some of the limitations. I then discuss how this relatively new experimental capability may unfold in the near future, along with some implications for social and behavioral science. http://research.microsoft.com/apps/video/dl.aspx?id=178910
Modeling U s er Po s ting Behavior on S oci a l Media Zhiheng Zu (In s titute of Automation) Qing Yang (In s titute of Automation, CA S ) Friend or Frenemy? Predicting S igned Tie s in S oci a l Network s S huang-Hong Yang (Georgia In s titute of Technology) Alex S mola (Yahoo! Re s earch) Bo Long (Yahoo! Lab s ) Hongyuan Zha (Georgia In s titute of Technology) Yi Chang (Yahoo! Lab s ) S oci a l -Network Analy s i s U s ing Topic Model Youngchul Cha (UCLA) Junghoo Cho (UCLA) Exploring S oci a l Influence for Recommendation - A Generative Model Approach Mao Ye (Penn S tate Univer s ity) Xingjie Liu (Penn S tate Univer s ity) Wang-Chien Lee (The Penn s ylvania S tate Univer s ity)
许多公共健康的领导人物现在相信,日益发展的社会网络学能够用于改善健康和幸福感。(贾斯汀·伦特里亚(Justin Renteria)/供洛杉矶时报) By Eric Jaffe, Special to the Los Angeles Times 作者:埃里克·杰斐,洛杉矶时报专稿 September 13, 2010 2010年9月13日 The old folk concept that our personal health behaviors rub off on those around us has received a staggering amount of scientific support of late. Over the last few years, study after study has shown that weight gain, drug and alcohol use, even loneliness and depression aren't islands unto themselves but are powerfully contagious — capable of spreading within our social networks just as germs scatter after a sneeze. 近朱者赤近墨者黑,我们个人的健康行为会受到周边的影响这个旧观念得到了最近数量庞大的科学支持。近些年来,一项又一项的研究表明,超重、吸毒还有饮酒,甚至包括孤独和抑郁都不是孤立地发生在个人身上的,而是会严重地传染,正如打个喷嚏会导致病毒扩散那样,它们也会在我们的社会网络内部传播开来。 If your friends are smokers, you tend to light up too, studies show. If they're overweight, then your belt also feels a bit tight. If they're happy, chances are you're smiling too. And on and on. 研究显示,如果你的朋友是吸烟者,你可能也会抽上一两根。如果他们超重,那么你也可能会感觉到安全带有点紧。如果他们高兴,你也有机会把微笑挂在脸上。诸如此类。 Many public health leaders now believe this growing science of social networks can be used to improve health and well-being on a broad, population-sized scale. Some see the approach as a promising new front against the day's most urgent health problems, such as obesity, smoking and suicide. 许多公共健康的领导人现在认为,不断发展的社会网络学可用于改善广大的、规模人口的健康和幸福感上。其中有些人还把它视为对抗当今最为迫切的健康问题,如肥胖、抽烟和自杀的一个前景光明的新阵线。 Get breaking news alerts delivered to your mobile phone. Text BREAKING to 52669. "We've come to realize more and more that how people live and function in social networks is really important to health," says Deborah Olster, acting director of the Office of Behavioral and Social Sciences Research at the National Institutes of Health. What's less clear, at least so far, is the best way to nudge people toward healthy habits and away from destructive ones. Results from experiments are mixed — some efforts work, others don't. In March, the NIH issued a funding opportunity for scientists studying how to improve public health through social networks. “我们越来越意识到,人们在社会网络如何生活和行使职能对于健康来说真的非常重要”,美国国立卫生研究院行为与社会科学研究中心办公室主任德博拉·奥尔斯特(Deborah Olster)如是说。这个东西尚未弄清楚,但起码到目前为止,它是推动人们形成健康习惯、远离恶习的最好方式。实验结果比较复杂,有些有效,有些则不起作用。今年三月,国立卫生研究院(NIH)为科学家研究如何通过社会网络来改善公共健康提供了一个资助的机会。 Public health programs could tap into social networks in two main ways, says Dr. Nicholas Christakis of Harvard Medical School, co-author with James Fowler of UC San Diego of the 2009 book "Connected: The Surprising Power of Our Social Networks and How They Shape Our Lives." 哈佛医学院的尼古拉斯·克里斯塔克斯(Nicholas Christakis)博士与加州大学圣地亚哥分校的詹姆斯·福勒(James Fowler)是2009出版的《联系:我们社会网络的神奇力量及其如何重塑我们的生活》一书的联合作者,他说公共保健计划进军社会网络有两种主要途径。 One approach, says Christakis, is to manipulate the network of connections people have. Artificial social groups — think Weight Watchers — could be created to urge unhealthy members toward more healthful behaviors. 一种方式,克里斯塔克斯说,是操纵人们拥有的关系网络。人为的社会团体,比如说减肥中心,可以被建立起来去敦促不健康的成员往更健康的行为而努力。 A second approach, which Christakis considers far more promising, is to manipulate existing networks so that positive health messages become "contagious." In this model, public health programs would target key members of a social group in an effort to influence the network at large. 第二种办法,在克里斯塔克斯看来更有前途,就是操纵现有的网络以便让积极的健康信息“蔓延”起来。在这种模式里,公共健康方案会把社会团体中的关键成员作为目标,让其最大限度地影响到整个圈子。 When social scientists talk of "networks," they mean any social circle in which people trade ideas, share experiences and generally touch each other's lives — neighborhoods, offices, classrooms or online communities such as Facebook. 社会学家讨论“网络”的时候,他们指的是任何人们交换思想、分享经验以及普遍相互接触彼此生活的社交圈—邻里、办公室、教室或在线社区,如Facebook。 Identifying the most influential or popular people within these networks has become easy, Christakis says, due to advances in data analysis. The trick is to identify which members could give positive health efforts the best bang for the buck and to create the best design for such programs. 由于数据分析方面的进展,克里斯塔克斯说,识别出这些网络中最有影响力或最具人气的人已经变得很简单。其诀窍是识别出哪一个成员最能够货真价实地做出健康的努力,并能够为此类计划做出最佳的设计。 In a study published in 2003 in the American Journal of Public Health, led by USC professor of preventive medicine Thomas Valente, sixth-graders participated in an eight-week smoking-prevention program. The intervention was taught to groups of students (the "networks," in this case) by class-nominated peers, teacher-nominated peers or random peers. All told, class-nominated peer leaders conducted the program most effectively, with students in these networks reporting less intention to smoke and lower smoking rates one year later. 2003年在《美国公共卫生杂志》上发表了一项研究,该项研究由南加州大学的预防医学教授托马斯·瓦伦特(Tomas Valente)领导,由6年级的学生参与,进行了为期八个星期预防吸烟计划。干预由指定的同龄同学、同辈老师或随机的同龄人传授给各组学生(本例中即是“网络”)。总而言之,指定班级的同辈领导执行计划最有效,据统计,一年后,这些网络的学生吸引的意向和吸引率都更低。 But a follow-up study, also led by Valente, had less success. This time, Southern California high schoolers took part in a 12-lesson drug-abuse intervention. The program had a greater effect on health behavior when led by an influential, student-nominated peer than when led by a teacher — but with a caveat. Drug use declined only in relatively drug-free social circles; for those whose friends included drug users, the program actually elevated drug use, the researchers reported in the journal Addiction in 2007. 不过,一项后续研究,也是由瓦伦特领导的,却没那么成功。这一次,南加州的高中生参与了一项12个课程的药物滥用干预。如果小组是由一个有影响的、同龄学生领导的,相对于由老师领导而言,该计划的效果更好,不过,有个警告。毒品使用只在相对远离毒品的社交圈子出现下降;对于那些朋友中有人吸毒者的人来说,实际上该计划反而会提高其对毒品的使用,这是研究人员在2007年的《成瘾》杂志上的报告。 The mixed findings suggest to Valente, who's done some of the leading work on network-based health interventions, that programs may need different designs based on the type of social network they're intended to help. This important point, he adds, has been entirely ignored in all work on health promotion. 瓦伦特进行了部分以网络为基础的健康干预的领导工作。这份结论不一的发现建议瓦伦特该计划可能需要根据他们希望帮助的社会网络的类型不同进行不同的设计。这个重点,在整个健康促进工作中却完全被忽略了。 "You can't divorce the content of the program from the people delivering it," he says. "The message is really the messenger." “你不能把计划的内容跟执行计划的人分隔开来”,他说:“信息真的就是信使”。 Valente is part of a new effort, led by Peter Wyman of the University of Rochester, to study how effectively key network members can implement a suicide prevention program called Sources of Strength. The intervention familiarizes students with suicide-coping resources and encourages them to seek help for suicidal friends. Peer leaders at 36 high schools in New York and North Dakota will learn the program and then introduce it into their social circle — the "network." Over the course of the five-year study, the researchers expect that students in these networks will seek more help than students in control groups, resulting in fewer suicide attempts. 瓦伦特是由罗切斯特大学的彼得·怀曼领导一个新举措的一部分,研究关键的网络成员实施自杀干预计划的有效性。干预行动让学生熟悉自杀应对的资源,并鼓励他们为有自杀倾向的朋友寻求帮助。纽约和北达科塔州36所高中的伙伴领导将学习该计划,然后把它介绍给自己的社交圈子—即“网络”。在为期5年的研究里,研究者预期这些网络中的学生将会比受控组的学生得到更多的帮助,其结果是自杀企图会更少。 Network interventions might be effective against obesity too, says economist Scott Carrell of UC Davis. In a recent study, Carrell and collaborators examined the spread of fitness habits in students at the Air Force Academy. The researchers found that the probability of a student being classified "unfit" tripled once half the student's social network fell out of shape. 网络干预对于防止肥胖也可能有效,加州大学戴维斯分校的经济学家挨斯科特·卡雷尔(Scott Carrell)说。在最近的一项研究里,卡雷尔和合作者研究了空军学院学生健身习惯的传播。研究者发现,一个学生被列为“不健康”的几率比身材走样的学生的社会网络的要高出两倍,但这个数字曾是他们的一半。 The finding underscores that network influences work in both directions — for good or for bad — but Carrell sees an opportunity in the results. "If you make that 'unfit' person more healthy," he says, "that suggests you will also increase the fitness levels of everyone else in the group." 这一发现强调了网络影响是双向的,好坏皆有,但卡雷尔从中看到了机会。“如果你让‘不健康’的人更健康”,他说:“那么这也意味着你将提高其所在团体的其他人的健康水平”。 All these efforts depend on a key scientific concern: to craft strong social-network interventions, researchers must first be certain that one person's health behavior has a direct, causal effect on another's, says Jason Fletcher of Yale University's School of Public Health. For some behaviors, such as smoking or drinking, the direct network effect is quite apparent, Fletcher says. But for others, such as obesity, the case is less clear. 一切努力都取决于一个关键的科学关注点:为了打造出强大的社会网络干预,研究者必须首先确定一个人的健康行为对他人的行为有着直接的、有因果关系的影响,耶鲁大学公共卫生学院的贾森·弗莱彻(Jason Fletcher)说。对于一些行为如吸烟或饮酒来说,直接的网络效应相当明显,弗莱彻说。但对于其他而言,如肥胖,情况就不是很明朗。 In a 2007 issue of the New England Journal of Medicine, Christakis and Fowler reported that obesity spread through social ties in one's neighborhood. They analyzed a real-world network population of more than 12,000 people living in Framingham, Mass., who were followed for 32 years, and concluded that overweight people tended to create overweight friends. 在2007年发行的《新英格兰医学》杂志里,克里斯塔克斯和福勒报告提到说,肥胖通过一个人的邻里关系传播。他们分析了现实世界里居住在马萨诸塞州弗雷明汉的超过12000个的网络人口,一直跟踪了32年,他们得出的结论是超重者倾向于结交的朋友也是超重的。 That finding, though intriguing, doesn't prove cause and effect, Fletcher argues. Environmental factors could have been a big influence: Maybe two neighbors are obese not as a result of their social connection but because a fast-food chain opened at the end of their street. Friend selection matters too: Maybe overweight people seek each other and cluster together rather than directly influencing one another's behavior. (Christakis says his study did consider these factors.) 这一发现虽然很有趣,但并未能证明因果关系,弗莱彻认为。环境因素也会是一个很大的影响:两个邻居肥胖的原因也许并不是因为其社会关系,而是由于在其所在街道头头开张的那个快餐连锁店。择友也会有影响:可能超重者互相寻找并聚在一起,而不是直接相互影响对方行为。(克里斯塔克斯说他的研究的确考虑了这些因素)。 A new study, presented in June at a conference of the American Society of Health Economists, tried to tackle this cause-and-effect dilemma by studying college roommate assignments — which are random in terms of mental health. The researchers, led by Daniel Eisenberg of the University of Michigan, then tracked the health behaviors of the 775 male and 867 female participants. 在今年6月举行的一次美国健康经济学家学会的会议上,发表了一份新研究,该研究试图通过研究大学划分宿舍来解决这一因果关系难题,这个分配在心理健康方面是随机进行的。由密歇根大学的丹尼尔·艾森伯格领导的研究人员,随后跟踪了775名男性和867名女性参与者的健康行为。 They found preliminary evidence for depression contagion only in male networks and for anxiety contagion only in female groups. Additionally, Eisenberg found plenty of support for the spread of binge drinking, but none for happiness. 他们发现了抑郁情绪仅在男性网络中传染,而焦虑情绪仅在女性群体中传染的初步证据。此外,艾森伯格发现了酗酒传播的大量支持,但幸福感却没有找到传播的依据。 Such a finding could help public health leaders build depression programs specifically for college-age males, Eisenberg says, but does throw some cold water on the notion that hanging around happy people is bound to make you more happy. 这样一个发现能帮助公共健康的领导针对大学年龄的男性建立抑郁控制的计划,艾森伯格说,但这也给那种认为围着快乐的人转能让你更快乐的观念泼了一点冷水。 Ultimately, experts agree, as understanding about the structure and flow of social networks improves, so will health interventions. Sorting out the details is a matter for science — for the average person, it's enough to know that improving the health of loved ones often means leading by example. 最后,专家同意,随着对社会网络的的结构和流动的理解加深,健康干预也会随之改善。理顺这些细节是科学的任务,对于普通大众来说,知道要想改善所爱之人的健康状况通常意味着要以身作则这一点就足够了。 "An easy thing to say would be 'Choose your friends wisely,' " says Olga Yakusheva of Marquette University, who has found that weight-loss behaviors can be contagious among college women. "I'd like to say, 'Choose your behaviors wisely,' because they're going to impact people around you." 马凯特大学的奥尔加·亚库谢娃(Olga Yakusheva)发现了减肥行为会在大学的女性同学中传播。“简而言之,就是要‘慎择良友’”,她说:“而我想说的是,‘小心行事’,因为这会影响到你周围的人”。 原文: http://article.yeeyan.org/view/boxi/135835
社会学中的社会网络 近年来,社会网络成为计算机领域的热门研究话题。作为一个研究分支,社会网络在社会学中的发展已超过了半个世纪,形成了一套比较有效的概念体系和研究方法,对当前计算机领域内的社会网络研究应该有可借鉴之处。本文简要介绍社会学中的社会网络研究( social network analysis , SNA ),希望对计算机专业研究人员有所启发。 社会网络研究介绍 传统的定量社会科学把个人的一些“标签”式的属性,如性别、收入、社会地位、阶级等,作为基本的分析单位,得到诸如性别比、人口统计、平均收入等指标,并研究其相互关系。以研究社会中的不平等现象为例,其标准的过程是:根据收入、职业等指标,对个人的社会地位进行量化,进而对量化结果进行统计分析,计算诸如均值、方差等参数,并试图建立其与性别、地域、受教育程度等因素的函数关系,再试图通过经济、文化、历史、社会心理来理解这些关系(现象)的成因。 不过,这种方法忽视了个人之间的社会交往对这些属性的影响。如统计平均收入,其实假定了个体的独立性。而正如俗语所言,“人以群分”,个人收入与朋友收入往往呈现正相关,并且个人往往会有意识地利用社会关系,来改善自己的社会地位。因此,属性化的分析多是一种“后观”式的描述,无法为解释社会现象提供系统的方法。 社会网络研究则是把关系放在中心的地位。在这套理论中,个人被抽象为节点,个人之间的社会关系作为节点之间的边,共同形成一个网络。社会学家希望网络的结构属性可以为社会现象提供系统性的解释。 相关研究内容包括: 个人的权力和声望 通过在网络中定义节点的度数,介数( betweenness )和接近度( close-ness )等概念,可以分别揭示个人在社会中声望某个方面的状况。如节点度数代表与一个人有关系的人数的多少;介数反映个人在网络中是否占据中间地位,隐含着沟通不同群体的能力;接近度则反映一个人与其他所有人的平均距离。在社会学意义下,这些概念蕴含着个人的权力或社会声望,反映一个人的社会资本。而社会网络中节点度数的分布则反映社会的分层情况 。 社会中的横向结构(社会群体) 社会中有不同的小群体,群体内部成员间的互相联系要比其与外部联系更强。社会网络研究给出了一些衡量方法,如 k- 派系( k-clique )和 k- 核( k-core )。计算机学科中对网络社群的研究很多就源于这些概念。 社会中的纵向结构(社会地位和角色) 社会学家引入了结构均衡( structural equivalence )的概念,用以分析个人的社会地位。例如,如果两个人在社会中具有几乎相同的社会关系人,则很大程度上两人在社会中有相同的作用,某种程度上可以被“互换”。进而可以把社会地位类似的节点划分到不同的类中,这也被称为块模型( blockmodel )。 社会的稳定与演变( structural balance ) 网络中不仅有朋友等互助关系,社会学家定义了关系的正负,正关系代表积极关系,如朋友,负关系代表某种敌意的关系。结构平衡用来衡量一个包含正负关系的网络是否稳定,并讨论不同情形下关系稳定的条件 。 对关系本身的研究 如社会网络的平均距离可以反映社会中信息传递的速度;聚簇因数反映关系的传递性,即个人与其朋友的朋友发生社会联系的可能性;密度则反映社会交往的频繁程度。 结合节点属性的研究 社会网络中的个人依然有其社会属性,如年龄、收入等信息。有些属性在社会网络的背景下进行统计,能获得新的认识,如判断社会关系是否促进经济收入提升等。 由上述内容我们能看出社会网络研究为经典的社会问题提供了一种新的阐释方式。 社会网络在社会学中的思想渊源 社会网络在社会学中的思想渊源可追溯到二战前 。 20 世纪 50 年代,二元组、三元组、结构平衡等 重要概念被提出。 20 世纪 70 年代后,计算机的应用让学者可以方便地分析收集到的社会网络数据,验证其理论。社会学家陆续提出了一些全新的观点,如著名的小世界现象。同时,一系列专门为社会网络分析设计的软件应运而生,如 UCINET 等。社会网络研究者还成立了自己的学术团体( INSNA )。沃瑟曼( Wasserman )和福斯特( Faust )于 1994 年出版了近千页的《社会网络分析:模型和方法》( Social Network Analysis:Model and Methods )可以作为经典社会网络研究的百科全书 。 社会网络研究的思想也扩展到了社会科学的其他领域。社会网络促进了经济社会学的新发展 ,网络与博弈论和市场研究的结合同样是近年来的研究热点 。而作为信息传播的依托,网络也成为传播理论的要素。在历史学研究中,社会网络研究被用来分析政治家族的派系关系 。一个有趣的例子是科林斯( Collins )研究了哲学家之间的学术网络,并认为这对哲学的发展有至关重要的影响 。 试举一个社会网络研究的经典例子。格兰诺维特( Granovetter )在 20 世纪 70 年代做的社会调查发现,在波士顿地区的职业技术人员中,那些通过个人关系找到工作的人,只有 16.7% 是依靠关系亲密的人(如密友、家人),而大多数人都是靠那些只有一面之交的人介绍的工作 。作者为了解释这种现象,引入网络模型,给边赋予强弱两种类型,并结合图论中捷径( local bridge )的概念,证明在强三元组假设下(即如果 AB 、 AC 有强关系,则 BC 至少有弱关系),起到捷径作用的边一定是弱关系。作者同时把这种思想延伸到了经济领域。经典经济学把个人视为理性、独立的行为人,追求最大利益。在这种理论下,密友等强关系多为个人有意经营的一种社会关系,需要时间和经济的投入,并且也随之能获得更大的回报。但是弱关系在职场上发挥的关键作用却与这一前提相悖,因此格兰诺维特把弱关系概念扩展到对市场行为的分析,冲击了理性人的假设,促进了社会学与经济的结合,对经济学的发展带来了某种范式型的转变 。这项研究从实证资料出发,根据现象提炼出规律,抽象出本质特征,得出数学化的结论,同时揭示其深刻的社会和经济学内涵,因此其相应论文成为有史以来被引用次数最多的社会学文献之一。 社会网络研究在中国有其特殊的意义。中国历来被认为是一个“关系”社会。梁漱溟提出的“关系本位”,费孝通倡导的“差序格局”,都以关系为中国社会的基础。近年来也有专门论述关系在现代社会中作用的著作 ,但这些研究都是集中在对关系的定性阐述。而随着社会网络研究影响的逐渐扩大, 2002 年出版的《中国的社会关系:制度、文化和关系本质的改变》( Social connections in China:Institutions, Culture, and the Changing Nature of Guanxi )一书中,谈到了不少学者使用社会网络研究的定量化研究方法,最后的总结还专门探讨了利用社会网络研究深入探索关系社会的可能性。这反映了社会网络研究的方法对传统课题研究的渗透 。 社会网络思想的逐渐普及使得“关系社会学”应运而生 。马克思的社会学理论更重视宏观社会结构对人类的决定性,而由韦伯和舒茨发端的现象学社会学则重视个人的自由意志对社会结构的影响。但宏观结构与微观个性之间的鸿沟在研究中长期存在。而社会网络研究的发展则提供了建立一种中层理论的可能性。相比微观的个人和宏观的社会,每个人生活的社会网络都是一种中层结构。它们影响微观的个人行为,而这些网络累加的效果又反映宏观社会现象。因此,基于网络或关系的社会学被视为跨越宏观与微观间鸿沟的桥梁 ,这可能为中国学者提供一个契机,真正发展出一种基于自己民族文化的社会思想。 社会网络研究对计算机学者的启发 通过简要回顾社会网络研究发展史,我们能发现,一些在计算机科学(应用图论)中的概念,如中心度,聚簇因数等,其源头都来自社会学。在互联网兴起之后,社会网络研究依然对计算机学科有所裨益。试举一例:社会学家弗里曼( Freeman )在 1977 年就提出了点介数的概念,反映一个节点在社会网络中处于中间人地位的程度,这逐渐成为衡量个人社会资本的一个标准 。而 2002 年,计算机科学家格文( Girvan )和纽曼( Newman )受到这个概念的启发,对点介数进行自然的扩展,利用边介数( edge betweenness )的概念,提出了 GN 算法 。这种方法能避免聚类算法的许多缺陷,是网络社区发现的经典算法之一。 近十年来,社会学家也尝试建立更加深入的网络模型。指数随机图模型( exponential randomgraph model , ERGM )即为其中一例 。如果说经典的社会网络研究更多是统计性描述,小世界网络等网络模型则是希望通过参数的设计,逼近现实的社会网络,而指数随机图则采取贝叶斯参数估计的方法,从实际数据中计算参数。我们可以根据自己的需要,计算网络中不同参数的重要性,如传递性、互惠性、中心性的强弱区分。根据不同的网络特性,可以加深对网络性质的了解。如研究链接预测时,可以根据网络的特性,决定具体的算法。网络可以通过聚簇因数预测传递性的强弱,中心性强的网络则可通过节点与网络中心的联系强度进行分析。 社会学与计算机科学在社会网络研究上最大的区别可能在于其讨论问题的视角。计算机领域更重视对模型的研究,通过设计模型,解决链接预测、社群发现等问题;而社会学更偏向经验层面,即通过网络进行统计分析,对网络的社会属性做出诠释。如对关系性质的探讨,像强弱、正负、互惠性(即两人之间关系是单向还是双向)、传递性( A 是 B 的朋友, B 是 C 的朋友,则 B 是否会促进 A 认识 C )等。 社会关系不是简单地用两点之间的边就可以描述的,对其社会意义的理解有助于深入研究。现有的很多模型都基于对网络的常识性认识,如富者愈富等。毕竟随着时间推移,对互联网的研究势必会细化,这就需要对网络的社会层面有更深入的认识。以前文提到的弱关系理论为例,计算机研究者已经开始意识到把社交网络的关系进一步细分的必要性 。这区别于现有社交网络中简单的好友 / 陌生人的设定,而受到社会学启发,互惠性也已经在研究中有体现 。 需要特别指出的是,计算机科学中的社会网络研究将互联网视为天然介质,而从社会网络研究发展历程可看出,互联网的兴起对社会学研究的变化起到了催化剂的作用,但它并不是一个必须条件。经典的社会学研究,如人口调查等,目前还无法用网络研究代替。如果我们能把世界划分为线上和线下,那么计算机学者的研究集中在互联网的框架之内,属于线上;而线下的社会行为则是经典社会学的研究范围。这两者之间的互动,如网络行为与实际生活的社会行为如何互相影响,在社会学家看来还需要一些社会调查数据的支持,如埃里森( El-lison )等对脸谱( Facebook )在生活中的影响所做的研究就结合了社会调查的方法 。仅通过对互联网分析得到的结论,虽然可以代表互联网本的性质,但在社会学家看来,不能简单地将其视为实际生活的代表,因为互联网反映的只是生活的一个侧面。 张 涵 北京大学计算机系和历史系本科生。主要研究方向为社会网络及其在信息科学、社会学中的应用,自然语言处理。 参考文献 S. Wasserman, Social network analysis: Methods andapplications. Cambridge University Press, 1994 D. Cartwright and F. Harary, Structural balance: ageneralization of Heider’s theory, Psychologicalreview, vol. 63, no. 5, 1956, 277~293 L. Freeman, The development of social networkanalysis. Vancouver: Empirical Press, 2004 M. S. Granovetter, Problems of explanation in economicsociology in Networks and Organizations: Structure,Form, and Action (N. Nohria and R. Eccles, eds.),Harvard University School Press, 1992, 25~56 M. Jackson, Social and economic networks. PrincetonUniversity Press, 2008 J. Padgett and C. Ansell, Robust action and the rise ofthe Medici, 1400-1434, American Journal of Sociology,vol. 98, no. 6, 1993, 1259~1319 R. Collins, The sociology of philosophies: a globaltheory of intellectual change. Belknap Press of HarvardUniversity Press, 1998 M. Granovetter, The strength of weak ties, Americanjournal of sociology, vol. 78, no. 6, 1973, 1360~1380 M. Granovetter, Economic action and social structure:The problem of embeddedness, American Journal ofSociology, vol. 91, no. 3, 1985, 481~510 M. Yang, Gifts, favors, and banquets: The art of socialrelationships in China. Cornell University Press, 1994 T. Gold and D. Guthrie, Social connections in China:Institutions, culture, and the changing nature of Guanxi,Cambridge University Press, 2002 M. Emirbayer, Manifesto for a relational sociology,American Journal of Sociology, vol. 103, no. 2, 1997,281~317 A. Mische, Relational sociology, culture, and agency,in The SAGE Handbook of Social Network Analysis(J. Scott and P. Carrington, eds.), pp. 80–98, SagePublications Ltd, 2011 L. Freeman, A set of measures of centrality based onbetweenness, Sociometry, vol. 40, no. 1, 1977, 35~41 M. Girvan and M. Newman, Community structurein social and biological networks, Proceedings of theNational Academy of Sciences, vol. 99, no. 12, p. 7821,2002 S. Wasserman and P. Pattison, Logit models and logisticregressions for social networks: I. an introduction toMarkov graphs and p* models, Psychometrika, vol. 61,no. 3, 1996, 401~425 E. Gilbert and K. Karahalios, “Predicting tie strengthwith social media,” in Computer Human Interaction,2009, 211~220 H. Kwak, C. Lee, H. Park, and S. Moon, What is twitter,a social network or a news media?, in Proceedings of the19th international conference on World wide web, ACM,2010, 591~600 N. Ellison, C. Steinfield, and C. Lampe, The benefits ofFacebook friends: social capital and college students’use of online social network sites, Journal of Computer-Mediated Communication, vol.12, no.4,2007,1143~1168 转自:中国计算机学会通讯,第 8 卷 第 6 期 2012 年 6 月 注:本文前两部分介绍了社会学社会网络研究的一些基本知识,第三部分的内容值得了解一下。
整理了一下WWW2012上有关社会网络的论文,主要是看文章名字和作者挑出来的,还没有细看,可能有一些是无关的,还有一些遗珠,不过足够看上一阵子了。 Actions speak as loud as words: Predicting relationships from social behavior data Sibel Adali, Fred Sisenda and Malik Magdon-Ismail Analyzing Spammers’ Social Networks For Fun and Profit — A Case Study of Cyber Criminal Ecosystem on Twitter Chao Yang, Robert Harkreader, Jialong Zhang, Suengwon Shin and Guofei Gu Bimodal Invitation-Navigation Fair Bets Model for Authority Identification in a Social Network Suratna Budalakoti and Ron Bekkerman Branded with a Scarlet ? C ?: Cheaters in a Gaming Social Network Jeremy Blackburn, Ramanuja Simha, Nicolas Kourtellis, Xiang Zuo, Matei Ripeanu, John Skvoretz and Adriana Iamnitchi Human Wayfinding in Information Networks Robert West and Jure Leskovec Implementing Optimal Outcomes in Social Computing: A Game-Theoretic Approach Arpita Ghosh and Patrick Hummel Information Transfer in Social Media Greg Ver Steeg and Aram Galstyan New Objective Functions for Social Collaborative Filtering Joseph Noel, Scott Sanner, Khoi-Nguyen Tran, Peter Christen, Lexing Xie, Edwin Bonilla and Ehsan Abbasnejad Online Team Formation in Social Networks Aris Anagnostopoulos, Luca Becchetti, Carlos Castillo, Aristides Gionis and Stefano Leonardi Partitioned Multi-Indexing: Algorithms, Analysis, and Applications to Social Search Bahman Bahmani and Ashish Goel Recommendations to Boost Content Spread in Social Networks Sayan Ranu, Vineet Chaoji, Rajeev Rastogi and Rushi Bhatt The Role of Social Networks in Information Diffusion Eytan Bakshy, Itamar Rosenn, Cameron Marlow and Lada Adamic Understanding and Combating Link Farming in the Twitter Social Network Saptarshi Ghosh, Bimal Viswanath, Farshad Kooti, Naveen Kumar Sharma, Korlam Gautam, Fabricio Benevenuto, Niloy Ganguly and Krishna Gummadi Using Content and Interactions for Discovering Communities in Social Networks Mrinmaya Sachan, Danish Contractor, Tanveer Faruquie and L. V. Subramaniam An Exploration of Improving Collaborative Recommender Systems via User-Item Subgroups Bin Xu, Jiajun Bu, Chun Chen and Deng Cai Community Detection in Incomplete Information Networks Wangqun Lin, Xiangnan Kong, Philip Yu, Quanyuan Wu, Yan Jia and Chuan Li Crosslingual Knowledge Linking Across Wiki Knowledge Bases Zhichun Wang, Juanzi Li, Zhigang Wang and Jie Tang Discovering Geographical Topics from Twitter Streams Liangjie Hong, Amr Ahmed, Siva Gurumurthy, Alex Smola and Kostas Tsioutsiouliklis Document Hierarchies from Text and Links Qirong Ho, Jacob Eisenstein and Eric Xing Dynamical Classes of Collective Attention in Twitter Janette Lehmann, Bruno Gon?alves, José Ramasco and Ciro Cattuto Factorizing YAGO: Scalable Machine Learning for Linked Data Maximilian Nickel and Volker Tresp Learning and Predicting Behavioral Dynamics on the Web Kira Radinsky, Krysta Svore, Susan Dumais, Jaime Teevan, Eric Horvitz and Alex Bocharov Vertex Collocation Profiles: Subgraph Counting for Link Analysis and Prediction Ryan N.Lichtenwalter and Nitesh V. Chawla We Know What @You #Tag: Does the Dual Role Affect Hashtag Adoption? Lei Yang, Tao Sun and Qiaozhu Mei