《大数据时代的网络科学.pptx》由会员分享，可在线阅读，更多相关《大数据时代的网络科学.pptx（45页珍藏版）》请在taowenge.com淘文阁网|工程机械CAD图纸|机械工程制图|CAD装配图下载|SolidWorks_CaTia_CAD_UG_PROE_设计图分享下载上搜索。
1、，Ocean University of China Big Data,Small WorldNetwork Science in the Data EraXiaofan WangCloud Atlas Cloud Atlas 云图云图Six connected stories Six hundred yearsHow lives are all linked Past,present and futureCloud AtlasCloud AtlasIts all about NETWORK!Various souls cross the ages like clouds cross the
2、skiesMessageMessageWe will never understand complex systems unlesswe gain a deep understanding of the networks behind themCommunicationA Network View of the WorldTransportationPower GridSocialBiologicalLanguageDifferent networks,similar featuresSchich,M.et al.Science 345(6196),558562(August,2014)A n
3、etwork framework of cultural historyBirth and death data of notable individuals reveal interactions between culturally relevant locations over two millennia.Milgram.Psychology Today,19671967,Milgram:Six-Degree Separation1967,Milgram:Six-Degree SeparationSmall World Feature721 721 million active mill
4、ion active users,4 users,4 degrees of separationdegrees of separationBackstrom et al.arxiv,2011Small World Facebook少量节点的度相对很大(Barabsi&Albert,Science,1999)Scale-free Feature通信网中的病毒传播电力网中的相继故障社会网中的病毒、信息传播经济网中的连锁反应传播行为网络传播r cActive phaseVirus deathn经典理论:c0网络传播传播临界值理论的修正n无标度网络：c=0 Pastor-Satorras,Phys.R
5、ev.Lett,2001R.Cohen et al.,PRL 91,247901(2003)随机熟人双熟人 目标 网络传播随机免疫、目标免疫、熟人免疫策略Justin Cheng,Lada A.Adamic,P.Alex Dow,Jon Kleinberg,Jure Leskovec,WWW,2014uIf we have seen k reshares,whats the prob.that a further k reshares will happen?uWhich is difficult?k=5 or k=100?Can Cascades be Predicted?网 络 科 学测量
6、网络数据发现网络性质建立网络模型分析网络行为设计网络性能Time-evolving Time-evolving InternetInternet1969 ARPANET1986 NSFNET1995 INTERNET1970 ARPANETHierarchicalHierarchical Internet InternetuIP leveluRouter leveluDomain(AS)levelMobile Communication NetworkMobile Communication NetworkGlobal Airline NetworkGlobal Airline Network
7、Networks that Networks that make the world more connectedmake the world more connectedInternetMobile CTransportationIs Is connection a basic human right?connection a basic human right?WWW&Search EngineWWW&Search EngineA great example of network thinkingWeb Viewed as a Directed NetworkWeb Viewed as a
8、 Directed NetworkNodes:Webpages(10 billion)Edges:Hyperlinks In-Degree of a WebpageIn-Degree of a WebpagePage B:In-D=2Page A:In-D=6nNumber of links point to the page nIs page A more important than page B?Yahoo!Your homepageMy homepageDefinition of PageRankDefinition of PageRankThe importance of a pag
9、e is given by the importance of the pages that link to itimportance of page ipages that link to page iout-degrees of page jimportance of page jPersonalized RecommendationBaseline:item popularityBaseline:item popularityDemographic-based RecommendationDemographic-based RecommendationContent based Reco
10、mmendationContent based RecommendationCollaborative FilteringCollaborative FilteringA Complex Network View of A Complex Network View of Recommender SystemsRecommender Systemsuu and v have a short path between them;uu and v have several paths between them;uu and v have paths that avoid high-degree no
11、des.Davidson,J.and Liebald,B.and Liu,J.The YouTube video recommendation system.Proceedings of the fourth ACM conference on Recommender systems.2010Baluja,S.and Seth,R.and Sivakumar,D.and Jing,Y.Video suggestion and discovery for youtube:taking random walks through the view graph.Proceeding of the 17
12、th international conference on World Wide Web.2008Recommender Systems&Link PredictionRecommender Systems&Link PredictionCommunicationNETWORK VIEW OF THE WORLDTransportationPower GridSocialBiologicalLanguage网网络络拓拓扑扑性性质质分分析析网络有多网络有多大大？连通性？连通性网络有多网络有多小小？平均距离？平均距离网络有多网络有多紧紧？聚类系数？聚类系数网络有多网络有多密密？连边密度？连边密度
13、网络有多网络有多匀匀？度分布？度分布网络有多网络有多配配？同配性？同配性网络有多网络有多分分？社团结构？社团结构应用研究应用研究如何刻画个体影响力？网络科学网络科学节点中心性Highest KatzHighest DegreeHighest BetweenessHighest ClosenessDiverse centrality measures应用研究应用研究如何从庞大的网络中找出特征群体？网络科学网络科学社区检测Network of NetworksNetwork of Networks(Interdependent Networks)Interdependent Networks)微博
14、人人Email微信QQ同一个人Lev Muchnik,Sinan Aral,and Sean J.Taylor,Social Influence Bias:A Randomized Experiment,Science 9 August 2013:341(6146),647-651.互联网时代的羊群效应互联网时代的羊群效应大众点评的餐厅评价大众点评的餐厅评价豆瓣的电影评价豆瓣的电影评价购物网站的商品评价购物网站的商品评价差差评师的影响有多大评师的影响有多大事先为101281篇网上文章随机给好评或差评，并与对照组相比事先给好评有引导性：读者给好评的可能性提高了32%事先给差评没有引导性：对文章最
15、后的评分几乎没有影响Romantic Partnerships and the Dispersion of Social Ties:Romantic Partnerships and the Dispersion of Social Ties:A Network Analysis of Relationship Status on FacebookA Network Analysis of Relationship Status on FacebookL.Backstrom,J.Kleinberg.Proc.17th ACM Conference on Computer Supported C
16、ooperative Work and Social Computing(CSCW),2014社社交交网网络络能能推推断断情情侣侣关关系系以以及及能否维持！能否维持！l l算算法法可可以以推推断断两两个个人人是是否否在在约约会会，准准确率确率60%60%l l如如果果算算法法没没有有检检测测到到，那那么么即即使使这这两两人人在在约约会会，两两个个月月内内分分手手的的概概率率为为50%50%1)构建网络（谁打给谁）2)研究细节（亲密程度）3)研究动态（网络演化）4)关联动态与突发事件网络科学的双刃剑网络科学的双刃剑u你的每一次通话或短信u你的每一封电子邮件u你在人人或微博上的每一次活动u你的每一次网页浏览、搜索u你的每一次网上或者现实购物u你在摄像头前的每一次出现数据时代：政府的边界在哪里？数据时代：企业的边界在哪里？The New Deal of Data The New Deal of Data 数据新政数据新政个人数据商店PDS你拥有关于你的数据的所有权使用你的数据的控制权关于你的数据的处置权网络科学、道德与法律科学家的奉献、企业家的良心法学家的智慧、政治家的责任公民意识的不断提高Complex Networks&Control Lab,SJTU