基于强弱联系反馈的多媒体社交网络信任模型

发布时间：2018-06-26 06:46

本文选题：多媒体社交网络 + 信任管理　；参考：《西安电子科技大学》2014年硕士论文

【摘要】：作为媒体分享技术和社交网络结合的新型网络形态,多媒体社交网络具有突出的社交特性和良好的传播特性。但由于多媒体社交网络的无中心架构特点以及缺乏动态灵活的管理机制,网络环境面临恶意内容泛滥,数据严重冗余等安全威胁。因此,如何在多媒体社交网络环境下建立一种信任管理机制,发现用户的恶意及共谋行为,限制社交网络中不良信息的传播,提高网络环境的安全性和维护网络的正常运行已经成为社交网络安全领域的一个重要研究方向。本文在研究多媒体社交网络所面临的安全威胁和分析社交网络经典信任模型的基础上,对多媒体社交网络环境下的信任管理进行了深入的研究。具体工作如下：1)提出了一种多媒体社交网络下的基于强弱联系反馈的信任模型。该信任模型由强弱联系的划分方案和基于强弱联系反馈的信任计算模型两部分组成。强弱联系的划分方案根据社会学中的弱联系理论和强联系的三度影响力准则,对多媒体社交网络中用户的身份进行划分。该划分方案分为个体划分和全网划分。在个体划分中,采用与目标用户相似度和目标用户的信誉值作为划分因素,对其邻居用户进行强弱联系的划分；全网划分算法是建立在个体划分基础之上,通过三度影响力准则,对全网络中的所有用户进行强弱联系身份的区分。在基于强弱联系反馈的信任计算方案中,根据强弱联系的划分方案结果,强联系和弱联系需要对目标用户进行评价值反馈：弱联系用户反馈内容评价值,强联系用户反馈内容评价值和行为评价值；利用上述评价值,信任模型采用李雅普诺夫中心极限定理和贝叶斯估计计算出该信任周期内强弱联系对目标用户的周期评价值；最后利用衰减窗口机制,结合近几个信任周期的周期评价值计算出当前信任周期内该用户的信誉值。通过仿真及模拟实验可以看出该模型具有合理性、有效性和健壮性,能够准确地、高效地反映出用户的信任情况,并能一定程度上有效抵制不良用户的恶意反馈和共谋攻击。2)提出了一种基于信任的多媒体社交网络的信息传播模型。首先依据用户在多媒体社交网络中对多媒体数字内容的具体操作,抽象出用户四个操作状态的转移图。然后采用经典SEIR传染病动力模型,构建基本的信息传播模型。再利用上述信任模型中的信誉值,通过社会学调查中信任与传播概率的趋势分析,研究在多媒体社交网络环境下信任管理方案对多媒体数据内容的传播所产生的影响。最后通过仿真平台的进行网络模拟实验得到,采用信任管理可以在很大程度上帮助促进健康数据在网络中的扩散,并抑制恶意数字内容的传播。
[Abstract]:As a new network form of media sharing technology and social network, multimedia social network has outstanding social characteristics and good communication characteristics. However, due to the characteristics of multimedia social network architecture and the lack of dynamic and flexible management mechanism, the network environment is faced with malicious content flooding, data redundancy and other security threats. Therefore, how to establish a trust management mechanism in the environment of multimedia social network to discover the malicious and collusive behavior of users, and limit the dissemination of bad information in social network. Improving the security of the network environment and maintaining the normal operation of the network has become an important research direction in the field of social network security. On the basis of studying the security threats faced by multimedia social networks and analyzing the classical trust models of social networks, this paper makes a deep research on trust management in multimedia social networks. The main work is as follows: 1) A trust model based on strong and weak link feedback is proposed for multimedia social networks. The trust model consists of two parts: the division scheme of strong and weak links and the trust computing model based on strong and weak link feedback. According to the weak connection theory in sociology and the three-degree influence criterion of strong connection, the user identity in multimedia social network is divided. The partition scheme is divided into individual partition and whole network partition. In the individual partition, the similarity with the target user and the credit value of the target user are used as the partition factors, and the neighbor user is divided into strong and weak connections. The whole network partition algorithm is based on the individual partition. Through the three degree influence criterion, all users in the whole network are distinguished between strong and weak contact identity. In the trust computing scheme based on strong and weak link feedback, according to the result of the classification scheme of strong and weak connection, the evaluation value feedback of the target user is needed: the content evaluation value of weak contact user feedback. Using the above evaluation value, the trust model uses Lyapunov's central limit theorem and Bayesian estimation to calculate the periodic evaluation value of the strong or weak connection to the target user in the trust cycle. Finally, the credit value of the user in the current trust cycle is calculated by using the attenuation window mechanism and combined with the periodic evaluation values of several trust cycles in recent years. Through simulation and simulation experiments, we can see that the model is reasonable, effective and robust, and can accurately and efficiently reflect the trust situation of users. To some extent, it can effectively resist malicious feedback and collusion attacks of undesirable users. (2) A model of information dissemination based on trust based multimedia social network is proposed. Firstly, according to the concrete operation of multimedia digital content in multimedia social network, the transfer diagram of user's four operation states is abstracted. Then the classical SEIR infectious disease dynamic model is used to construct the basic information transmission model. By using the credit value of the trust model and the trend analysis of trust and propagation probability in sociological investigation, the influence of trust management scheme on the spread of multimedia data content in multimedia social network environment is studied. Finally, through the network simulation experiment on the simulation platform, it is concluded that trust management can greatly help to promote the spread of health data in the network and suppress the spread of malicious digital content.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.08

【相似文献】