摘要
摘要
在互联网科技高速发展的时代，人们习惯于在互联网上进行信息的传递，庞大的用户体以及用户之间复杂的关系构成了现在丰富多彩的社交网络。社交网络平台上每天都会产生成千上万条热点话题，并通过各种途径在人中迅速的传播。因此，社交网络也成为了各种谣言话题滋生的摇篮。谣言在社交网络平台上传播的过程中，由于其本身的特殊性和社交网络平台用户数量的有限性，必然会受辟谣消息和衍生谣言的影响。如何有效的度量辟谣消息和衍生谣言对当前谣言的影响，以及对当前谣言传播趋势进行有效的预测是亟待研究和解决的问题。
本论文所研究的谣言传播模型主要从辟谣消息和衍生谣言消息两个方面出发，对社交网络中谣言的传播进行建模分析和研究。主要研究工作如下：1．在辟谣消息方面，充分考虑谣言与辟谣消息之间相互对抗竞争的关系，构造了基于演化博弈和辟谣消息的谣言传播模型。首先，针对谣言与辟谣消息的共生性，以传染病动力学中的SIR模型为基础，将辟谣消息作为一条独立的信息引入到谣言传播的过程中，构建了SKIR谣言传播模型。然后，针对谣言与辟谣消息的对抗特性以及人们的趋利性和从众心理，利用演化博弈理论，构建用户参与话题的驱动机制，探索谣言扩散过程中用户行为动力学成因。同时，结合用户个人行为特征、用户交互行为特征以及信息本身特性，利用多元线性回归方法构建用户行为驱动因素。最后，结合本文提出的SKIR模型，对在辟谣消息影响下谣言的传播情况进行建模分析。
2．在衍生谣言方面，考虑到当前谣言与衍生谣言之间相互竞争的关系，本文构建了基于Lotka-V olterra模型和衍生谣言的谣言传播模型。首先，针对社交网络中用户行为的复杂性，分别从用户自身行为和外部邻居两个方面，利用多元线性回归方法构建谣言的影响力。然后，考虑到衍生谣言与当前谣言的竞争性和谣言影响力的动态变化，利用Lotka-V olterra模型度量衍生谣言与当前谣言的竞争关系。最后，结合传统的SIR模型，构建状态转化方程，从而对衍生谣言影响下的谣言传播态势进行建模分析。
为了验证所提模型的有效性和准确性，本文采用新浪微博、腾讯微博以及网
重庆邮电大学硕士学位论文
络合成数据进行实验验证。实验表明，本论文所提出的模型能够有效的反应辟谣消息和衍生谣言对当前谣言的影响，并能对谣言传播趋势进行有效的预测。该模型有助于预测谣言在社交网络中的演化趋势和发展规律，为谣言的控制和多消息的传播提供理论基础。
关键词：社交网络，谣言传播动力学，演化博弈理论，Lotka-V olterra模型
Abstract
Abstract
In the era of Internet technology development, people are accustomed to communicating on the Internet. Large-scale users and intricate user relationships constitute a colorful online social network. Thousands of hot topics emerge and spread rapidly through various channels. Therefore, the social networking platform has become a paradise for all kinds of rumors. Due to the particularity of rumors and the limited number of users, the spread of rumors will be affected by anti-rumors and derivative rumors in social networks. How to effectively measure the impact of anti-rumors and derivative rumors on current rumors, as well as effective prediction of current rumor propagation trends, is an issue that needs to be studied and solved.
This thesis mainly analyzes and studies rumor’s propagation in social networks from two perspectives, namely anti-rumors and derivative rumors. The main research work can be summarized as follows:
1.In studying anti-rumor, we fully consider the relationship between rumors and anti-rumors, and construct a rumor propagation model based on evolutionary games and anti-rumors. Firstly, as for the symbiosis of rumors and anti-rumors, based on the SIR model of infectious disease dynamics, anti-rumors is introduced as an independent information into the spreading process of rumors to construct the SKIR model. Secondly, in view of the confrontational characteristics of rumors and anti-rumors, as well as people's profitability and herd mentality, the Evolutionary Game Theory is adopted in this paper
to construct the drive mechanism of user participation in the topic, and explores the causes of user behavior dynamics in the spreading process of rumors. At the same time, we combined internal factors and external factors of the users to build the influence of information by multivariate linear regression method, which provides the theoretical basis for user behavior driving force. Finally, we combined with the SKIR model proposed in this paper to model and analyze the propagation of rumors under the influence of anti-rumors.
2.For derivative rumors, considering the competing relationship between current rumors and derivative rumors, this paper builds a rumor propagation model based on Lotka-V olterra model and derived rumors. First of all, due to the complexity of user behavior in the social network platform, the multiple linear regression method is used to
重庆邮电大学硕士学位论文
construct the user's behavior driving force from the user's behavior and influence of neighbor node. Then, considering the competitiveness between derivative rumors and current rumors, and the dynamics of rumor influence, the Lotka-V olterra model is used to measure the competitive relationship between derived rumors and current rumors. Finally, we combine the traditional SIR model to construct
the state transformation equation and model the propagation situation of the current rumors under the influence of derived rumors.
In order to verify the validity and accuracy of the proposed model, this paper uses Sina Weibo, Tencent Weibo and network synthesis data for experimental verification. Experiments show that the model proposed in this paper can effectively reflect the influence of anti-rumors as well as derived rumors on current rumors, and can accurately predict the propaganda trend of rumors. This model helps in predicting the evolutionary trends and development laws of rumors in social networks, and at the same time provides a theoretical basis in internet rumors control as well as the dissemination of multiple messages.
Keywords:social networks, rumors propagation dynamics, evolutionary game theory, Lotka-V olterra model
网络谣言的危害目录
目录
图录 .............................................................................................................................. V III 表录 ................................................................................................................................ IX 第1章绪论 .. (1)
1.1课题研究背景及意义 (1)
1.2 研究现状 (2)
1.2.1 社交网络发展现状 (2)
1.2.2 谣言传播模型概述 (3)
1.3 论文主要内容与意义 (5)
1.4 论文的组织架构 (7)
第2章相关技术和理论基础 (8)
2.1 用户的影响力分析 (8)
2.2 信息传播模型 (9)
2.2.1 传染病模型 (9)
2.2.2 线性阈值模型 (11)
2.2.3 独立级联模型 (11)
2.3 演化博弈相关理论 (12)
2.4 Lotka-V olterra相关理论 (15)
2.4.1 Lotka-V olterra基本模型 (15)
2.4.2 Lotka-V olterra演变模型 (15)
2.5 基本复杂网络模型 (16)
2.6 本章小结 (19)
第3章基于演化博弈和辟谣消息的谣言传播模型 (20)
3.1 引言 (20)
3.2 特征提取与定义 (20)
3.2.1 用户内部行为因素 (20)
3.2.2 用户心理因素 (21)