社交网络上信息传播过程的相关研究

发布时间：2018-03-09 20:41

本文选题：社交网络　切入点：SDIR信息传播模型　出处：《兰州理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：近年来随着互联网技术的突飞猛进,在线社交网络正在逐步改变人们的信息获取方式与社交方式。以美国的Facebook、Twitter、Youtube,以及国内的新浪微博、QQ、微信等为代表的大量在线社交网络发展势头迅猛,已经累计了数以亿计的用户,影响力巨大。社交网络以其实时性、功能性、社交性等诸多优势成为web2.0体系结构下最重要的应用之一,对现在社会中各种新闻事件、不实谣言、群众舆论等信息的传播有重要影响,利用好社交网络这种新兴工具可以为人类创造极大的价值,但同时社交网络在信息传播上的优势也可能被不当利用而造成巨大的危害。因此以社交网络为研究对象的信息传播研究具有十分重要的意义,也是近年来社会计算领域的一个热点。目前社交网络方面的研究焦点之一是以经典的传染病动力学模型为基础,来挖掘特定网络上的信息传播规律。本文首先介绍了社交网络的发展现状与研究意义,陈列出研究所涉及的复杂网络与信息传播的基础知识;然后针对社交网络中信息传播的特点,在传统的SIR模型基础上,通过加入新的一类假免疫节点建立了SDIR模型;最后在此基础上考虑到邻居节点间的相互影响,定义了三个传播概率函数,对SDIR模型做了改进。通过对比不同条件下信息传播的过程,实验证明了信息不能覆盖全网络,且Twitter比新浪微博有更好的信息传播效率的推测,并发现初始传播概率会对信息传播有重要影响。在上述仿真模拟SDIR模型的过程中,必须提前以经验值设定基本的传播概率,才可以运用各种传播模型模拟信息的传播过程。然而本文通过分析发现人为设定的传播概率对精确描述传播过程有很大影响,故首先以知识图谱补全问题中路径排序算法的思想为基础,提出了一种通过随机游走来计算节点影响力的算法;然后在节点影响力的基础上通过归一化得到初始的信息传播概率。实验对比了固定传播概率与考虑了信息源节点影响力的传播概率对传播结果造成的差异,通过证明影响力算法的有效性,验证出计算后的传播概率更加符合真实情况,并将通过计算得到的传播概率与SDIR模型结合,进一步完善了本文提出的信息传播模型。
[Abstract]:In recent years, with the rapid development of Internet technology, Online social networks are gradually changing the way people access information and how they socialize. A large number of online social networks, such as Facebook Twitter Youtubein the United States, and Sina Weibo QQQ, WeChat in China, are developing rapidly, and have accumulated hundreds of millions of users. Social network has become one of the most important applications under the web2.0 architecture because of its real-time, functional and social advantages. The dissemination of public opinion and other information has an important impact. Making good use of social networks as a new tool can create great value for human beings. But at the same time, the advantage of social network in information dissemination may also be improperly used to cause great harm. Therefore, the study of information dissemination based on social network is of great significance. It is also a hot topic in the field of social computing in recent years. At present, one of the research focuses on social networks is based on the classical dynamics model of infectious diseases. Firstly, this paper introduces the development status and research significance of social network, and displays the basic knowledge of complex network and information dissemination. Then, according to the characteristics of information transmission in social networks, based on the traditional SIR model, the SDIR model is established by adding a new class of pseudo-immune nodes. Three propagation probability functions are defined, and the SDIR model is improved. By comparing the process of information transmission under different conditions, the experiment proves that the information can not cover the whole network, and Twitter has better information transmission efficiency than Sina Weibo. It is also found that the initial propagation probability has an important effect on the information transmission. In the process of simulating the SDIR model mentioned above, the basic propagation probability must be set in advance with the experience value. However, through the analysis, it is found that the artificially set propagation probability has a great influence on the accurate description of the propagation process. Therefore, based on the idea of path sorting algorithm in the knowledge map complement problem, this paper proposes an algorithm to calculate the influence of nodes by random walk. Then the initial information propagation probability is obtained by normalization on the basis of node influence, and the difference between the fixed propagation probability and the transmission probability considering the influence of the information source node is compared. By proving the validity of the influence algorithm, it is verified that the calculated propagation probability is more in line with the real situation, and the information propagation model proposed in this paper is further improved by combining the calculated propagation probability with the SDIR model.
【学位授予单位】：兰州理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：G206;TP393.09

【参考文献】