基于效用的社交网络用户信息隐私保护算法研究
发布时间：2018-12-13 17:10
【摘要】：移动互联网、社交网络愈发深刻地融入人们的日常生活,特别是在发展势头猛烈的物联网、多样化的个性化服务的相互作用下,越来越多的社交用户的隐私信息被主动或无意地暴露在网络环境中。另外,大量的社交网络数据,为网络应用提供商带来更多利益的同时,也为恶意攻击者提供了动机。因此,社交网络环境下,针对用户信息的隐私保护问题的研究具有重大的理论和现实意义。近年来,研究人员针对社交网络中的隐私保护问题,提出了多种匿名模型和算法,但是鉴于匿名算法为了实现隐私保护,需要对社交网络数据产生不同程度的干扰,从而降低服务提供商对用户数据挖掘分析的准确度,使用户的服务体验大打折扣。可见,隐私安全与数据的可用性是社交网络环境下隐私保护问题研究的两个相互矛盾的目标,研究隐私安全和可用性之间的权衡成为一个重要和具有挑战性的问题。另外,相对于信息隐私安全方面的研究,数据效用的研究尚不成熟,目前还没有标准的可供普遍使用的数据效用损失度量标准。因此,本文以社交网络中用户属性为研究对象,致力于研究在确保用户信息隐私安全的前提下能够有效降低数据效用损失的匿名模型和算法,研究内容主要包括以下三部分:(1)针对目前社交网络场景下的隐私保护问题,现有的匿名算法一般以边或节点的修改数量作为评估匿名数据效用损失的唯一标准,这些算法隐私保护程度较高,但是由于忽略了不同的边或节点的修改对社交网络结构具有不同的影响,容易导致过量的数据效用损失,影响匿名数据的利用价值。考虑到这一问题,本文从数据效用具有结构相似性和信息完整性两个方面的角度出发,设计了一种更全面的数据效用衡量方法UL(G,G’),该方法综合评估匿名操作对网络结构和数据内容两方面的影响,其评估标准较以往仅以人为更改操作数量衡量效用损失的标准更高,降低了匿名数据的效用损失。(2)为了解决社交用户信息隐私安全问题,我们改进k-度-l-多样性匿名模型,提出了基于节点分裂的差异化隐私保护匿名模型((d,k,l)-u匿名模型),并基于该模型设计了相应的属性差异化匿名算法,该算法根据敏感度函数,将敏感属性的属性值划分到高、中、低三个隐私匿名组中,并对不同的匿名组采用不同程度的匿名规则。该算法将隐私保护对象由属性类精确到具体的属性值,并通过差异化匿名降低了匿名数据的效用损失,并通过模拟实验,验证了该算法的有效性。(3)针对匿名算法对网络结构的干扰,考虑到社交网络中的用户节点对网络的整体结构具有不同的影响力,即相对于普通节点,对“桥节点”等关键节点的分裂操作,会大幅度更改网络结构的整体特性,因此,本文对(d,k,l)-u属性差异化算法进行优化,进而提出了一种节点差异化的匿名算法(DKDLD-U匿名算法),该算法引入社会网路分析中的关键节点分析,将节点分为重要节点和普通节点两类,并对两类节点分别采用敏感属性值泛化和节点分裂两种匿名操作,以减少对网络结构的扰动,提高发布数据的效用。算法的仿真实验表明,该算法能够在保证隐私安全的同时,有效降低匿名数据的效用损失。
[Abstract]:Mobile Internet and social networks are becoming more and more deeply integrated into people's daily lives, especially with the interaction of dynamic Internet of Things and diversified personalized services, and more and more social users' privacy information is actively or unintentionally exposed to the network environment. In addition, a large amount of social network data provides a motive for a malicious attacker while bringing more benefits to the network application provider. Therefore, in the social network environment, the research on the privacy protection of the user information has great theoretical and practical significance. In recent years, the researchers put forward a variety of anonymous models and algorithms for privacy protection in social networks, but in view of the anonymity algorithm in order to realize the privacy protection, it is necessary to generate different degree of interference to the social network data. so that the accuracy of the service provider to the user data mining analysis is reduced, and the service experience of the user is greatly reduced. It can be seen that the security of privacy and the availability of data are two conflicting goals of the study of privacy protection under the social network environment, and the trade-off between privacy security and availability becomes an important and challenging issue. In addition, the research on the safety of information privacy, the research of data utility is not mature, and there is no standard of data utility loss measurement standard that can be widely used at present. Therefore, on the premise of ensuring the privacy and security of the user's information, this paper is devoted to the study of the anonymous model and the algorithm which can effectively reduce the loss of data utility under the premise of ensuring the privacy and safety of the user, and the research contents mainly include the following three parts: (1) Aiming at the privacy protection problem in the current social network scene, the existing anonymous algorithm generally takes the modified quantity of an edge or a node as the only standard for evaluating the utility loss of the anonymous data, and the privacy protection degree of the algorithms is high, However, due to the fact that the different edges or the modification of the nodes have different influences on the social network structure, the excessive data utility loss can be easily caused, and the utilization value of the anonymous data is affected. In view of this problem, a more comprehensive method of data utility measurement (UL (G, G '), the method comprehensively evaluates the influence of the anonymous operation on the network structure and the data content, and the evaluation standard of the method is higher than that of the prior art only by man-made change operation quantity, and the utility loss of the anonymous data is reduced. (2) In order to solve the security problem of social user information privacy, we improve the k-degree-l-diversity anonymous model, and propose a differential privacy protection anonymous model based on node division ((d, k, l)-u anonymous model). and a corresponding attribute differentiation anonymity algorithm is designed based on the model, and the algorithm divides the attribute value of the sensitive attribute into the high, middle and low privacy anonymous groups according to the sensitivity function, and adopts the different degree of anonymous rules for different anonymous groups. In this algorithm, the privacy protection object is defined by the attribute class to the specific attribute value, and the utility loss of the anonymous data is reduced by the differential anonymity, and the validity of the algorithm is verified through the simulation experiment. (3) Aiming at the interference of the anonymous algorithm to the network structure, the user node in the social network has different influence on the whole structure of the network, that is, the split operation of the key nodes such as the 鈥渂ridge node鈥，
本文编号：2376895
[Abstract]:Mobile Internet and social networks are becoming more and more deeply integrated into people's daily lives, especially with the interaction of dynamic Internet of Things and diversified personalized services, and more and more social users' privacy information is actively or unintentionally exposed to the network environment. In addition, a large amount of social network data provides a motive for a malicious attacker while bringing more benefits to the network application provider. Therefore, in the social network environment, the research on the privacy protection of the user information has great theoretical and practical significance. In recent years, the researchers put forward a variety of anonymous models and algorithms for privacy protection in social networks, but in view of the anonymity algorithm in order to realize the privacy protection, it is necessary to generate different degree of interference to the social network data. so that the accuracy of the service provider to the user data mining analysis is reduced, and the service experience of the user is greatly reduced. It can be seen that the security of privacy and the availability of data are two conflicting goals of the study of privacy protection under the social network environment, and the trade-off between privacy security and availability becomes an important and challenging issue. In addition, the research on the safety of information privacy, the research of data utility is not mature, and there is no standard of data utility loss measurement standard that can be widely used at present. Therefore, on the premise of ensuring the privacy and security of the user's information, this paper is devoted to the study of the anonymous model and the algorithm which can effectively reduce the loss of data utility under the premise of ensuring the privacy and safety of the user, and the research contents mainly include the following three parts: (1) Aiming at the privacy protection problem in the current social network scene, the existing anonymous algorithm generally takes the modified quantity of an edge or a node as the only standard for evaluating the utility loss of the anonymous data, and the privacy protection degree of the algorithms is high, However, due to the fact that the different edges or the modification of the nodes have different influences on the social network structure, the excessive data utility loss can be easily caused, and the utilization value of the anonymous data is affected. In view of this problem, a more comprehensive method of data utility measurement (UL (G, G '), the method comprehensively evaluates the influence of the anonymous operation on the network structure and the data content, and the evaluation standard of the method is higher than that of the prior art only by man-made change operation quantity, and the utility loss of the anonymous data is reduced. (2) In order to solve the security problem of social user information privacy, we improve the k-degree-l-diversity anonymous model, and propose a differential privacy protection anonymous model based on node division ((d, k, l)-u anonymous model). and a corresponding attribute differentiation anonymity algorithm is designed based on the model, and the algorithm divides the attribute value of the sensitive attribute into the high, middle and low privacy anonymous groups according to the sensitivity function, and adopts the different degree of anonymous rules for different anonymous groups. In this algorithm, the privacy protection object is defined by the attribute class to the specific attribute value, and the utility loss of the anonymous data is reduced by the differential anonymity, and the validity of the algorithm is verified through the simulation experiment. (3) Aiming at the interference of the anonymous algorithm to the network structure, the user node in the social network has different influence on the whole structure of the network, that is, the split operation of the key nodes such as the 鈥渂ridge node鈥，
本文编号：2376895
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2376895.html
