在线社交网络中谣言阻断的关键节点挑选方法研究

【作者】 张敏；

【导师】 王小明；

【作者基本信息】 陕西师范大学 ， 计算机软件与理论， 2021， 硕士

【摘要】 随着科技的进步和互联网技术的飞速崛起,Twitter、Facebook以及新兴的Tik Tok、WeChat等在线社交网络(Online Social Networks,OSNs)迅速普及到广大人民群众中,为人们提供了交流互动、资源共享、在线购物等省时省力的功能,成为当今时代人们从事各种社会活动的主流媒介。此外,OSNs的发展也为信息学、心理学、社会学、计算机科学等领域的学者提供了新的研究方向,如信息传播模式、用户购物行为的心理倾向分析、公共突发事件管控、社区检测和数据挖掘与处理等等。因此,OSNs具有非常重要的研究意义和价值。然而,由于OSNs的开放性与便利性,谣言等不良信息也能够进行快速扩散,会严重影响社会秩序与社会稳定,例如2020年各种新冠病毒相关的谣言大大影响了国家疫情的防控和治理。因此对于谣言扩散过程和谣言阻断方法的研究已经成为社会各个领域的专家学者们共同关注的问题。事实上,由于OSNs拓扑结构的复杂性和OSNs的主体——人的复杂性,同时在谣言的扩散过程中也充满了很多不确定的因素,现有的关于谣言扩散过程以及谣言控制方法的研究的切入点都有所不同,如针对谣言特性、谣言数量、用户社区划分等等。本文考虑了网络拓扑特性以及用户特性,提出了两种方法来挑选并阻断谣言扩散过程中的关键节点以达到减小谣言不良影响的效果,具体内容如下:1)一种基于用户忍受度的阻断节点挑选方法。在现实生活中可以发现OSNs上的用户拥有如下特点:普通用户的关注用户数与被关注用户数之间的差距很小,而影响力大的用户其关注用户数与被关注用户数之间的差距则非常巨大。因此考虑到可以通过用户关注数与被关注数之间的差距来衡量节点的影响力的大小,提出了基于信息熵的影响力节点判定方法。此外,考虑到直接阻断网络中影响力大的用户可能会引起其被关注用户的不满并进一步引发更大的网络风波,文中提出了用户阻断忍受度的概念,当一个用户的影响力越大时,其用户阻断忍受度就越低;反之,当一个用户的影响力越小时,其用户阻断忍受度就越高。每个用户都会有一个忍受阈值,表示对于谣言阻断忍受的临界值,当其用户阻断忍受度小于该阈值时可以对其进行阻断,否则不能对其进行阻断。总的来说,本方法就是在用户的阻断忍受度之内挑选出关键节点进行阻断,以达到既不影响用户体验也能够起到阻断谣言扩散进程的效果。2)一种基于结构洞的阻断节点挑选方法。无论是积极有益的信息还是负面消极的信息如谣言,它们的扩散过程与网络拓扑结构密切相关。在我们看来,一个用户是否被激活与它的邻居节点有关,如果一个用户的邻居用户误信谣言并且该用户与之关系密切,那么他误信谣言的可能性非常大。因此,本文定义了一个新的概念——关系质量来衡量节点之间的亲密程度,并在此基础上构建了一个新的模型来描述谣言在OSNs中的扩散过程。此外,考虑到处于结构洞位置的节点在OSNs中颇具优势,这些节点在网络上拥有很大的“权力”或者说是影响力,能够对谣言的扩散过程产生很大的影响。因此我们引入了结构洞理论的四个测量指标,并对这些指标进行了分析。通过这四种指标选择有影响力的节点来控制谣言等不良信息进一步扩大其不良影响。最后,我们在扩散模型中采用了动态阻断方法以及单次阻断方法,并比较了采取不同阻断节点挑选策略时,网络中谣言感染率的变化情况。更多还原

【Abstract】 With the advancement of technology and the rapid rising of the Internet,Online Social Networks(OSNs)such as Twitter,Facebook and emerging Tik Tok have come to the broad masses of people.Providing with many convenient functions such as interaction,resource sharing and online shopping,OSNs have been the mainstream media for people to engage in various social activities in modern era.In addition,the development of OSNs has also brought new research subjects for scholars in Informatics,Psychology,Sociology,Computer Science and other fields.Specifically,information propagation mechanism,psychological tendency analysis of users’ shopping behavior,the management and control of public emergency,community detection and data mining and processing OSNs have drawn more and more attention.In summary,OSNs do have great research significance and value.However,due to the openness and the convenience of OSNs,rumors and other bad information can also spread rapidly,which tends to affect social order and social stability.For example,various rumors about COVID-19 in 2020 have seriously affected its prevention and control.Therefore,the study of diffusion of rumor diffusion and blocking has become a common concern of the whole society.In fact,as for the complexity of the topology structure of OSNs,the human beings and other uncertain factors in the process of rumor diffusion,the existing studies on these aspects have different breakthrough points.Considering the characteristics of network topology structure and the identities of OSNs,this paper proposes two methods to select and block the key nodes in the process of rumor diffusion in order to reduce the negative effects it brought.The concrete contents are as follows:1)A method for selecting key nodes based on user blocking tolerance.In real life,it can be found that the users on OSNs have an interesting common characteristic.For the ordinary users,there is a marginal difference between the number of their followers and the number of users followed by them.For the influential users,this difference is huge.Therefore,considering that the influence of users can be measured by the gap between the number of their followers and the number of users followed by them,the influential nodes selection method is proposed based on information entropy.In addition,out of the consideration that direct blocking of top influential users in the network may cause the dissatisfaction of their followers and further lead to greater network disturbances,the concept of user blocking tolerance is proposed in this paper and its value is inversely correlated with the influence of the users.Each user in OSNs will have a tolerance threshold,which represents the critical value of rumor blocking tolerance.It can be blocked only when its user blocking tolerance is not more than the threshold.In general,this method is to select the nodes to be blocked within the user’s tolerance for blocking,so as to block the rumor without affecting the user experience.2)A method for selecting blocking nodes based on structural holes.In fact,whether positive information or negative information such as rumors,their diffusion process is closely related to the network topology.In other words,whether a user is infected or not is related to its neighbors.If a user’s neighbor misbelieve the rumors and they have close relationship with each other,there will be a high probability for him believing in the rumors.Therefore,a new concept named relationship quality is defined to measure the degree of intimacy between nodes,and a new model to describe the rumor diffusion process in OSNs is built based on it.In addition,the nodes located in the structure holes have a great "power" or influence in the OSNs,which could play an essential part in the diffusion process of rumors.Therefore,four measurement indexes of structural hole theory are introduced and analyzed in this work.Through these four measurements,influential nodes are selected to blocking rumors for preventing their worse influence.Finally,the dynamic blocking method and the one-time blocking method are used to compare the changes of rumor infection ratio in the network after different blocking node selection strategies are adopted.更多还原