事务型数据流发布的隐私保护方法研究

发布时间：2018-08-07 10:53

【摘要】：近年来,随着网络信息技术的快速发展及其在人们生活中的应用,使得互联网中产生了大量的数据,这些数据包含大量信息,商业机构、政府部门可以利用这些数据来进行商业决策、科学研究等。因此数据共享变得非常有必要,然而这些数据包含了大量的个人敏感信息,如果数据拥有者发布这些数据前不做适当处理就有可能使得个体的隐私信息泄露,从而给个体带来危害。因此,数据发布中的隐私保护问题成为当前研究的热点。该问题研究如何对数据进行处理使得发布的数据不泄露个体隐私的同时具有较高的可用性。目前数据发布中的隐私保护技术大都是基于静态数据集的。随着物联网的推广以及大数据时代的到来,网络点击数据、电话呼叫记录、大型超市的购物数据等通常表现为动态变化的数据流,数据流具有海量性、实时性和动态变化性。传统的隐私保护技术不适用于数据流环境中。事务型数据流是数据流的一种普遍形式,如商场的购物数据等,其中每条记录是由项目组成的集合。这种数据通常包含用户的敏感信息。目前数据流发布中的隐私保护方法主要针对关系型数据流。事务型数据流具有高维和稀疏的特性,使得用于处理关系型数据流的隐私保护技术也不能直接适用于事务型数据流;事务型数据流具有实时和动态变化性,使得传统的处理静态事务型数据的隐私保护方法不能直接应用于数据流环境。本文研究事务型数据流发布的隐私保护问题,提出了基于滑动窗口的事务型数据流发布的隐私保护方法。主要研究工作如下:首先,结合滑动窗口和ρ-不确定性隐私保护模型,提出基于滑动窗口的ρ-不确定性模型,即要求任一滑动窗口都满足ρ-不确定性。当新数据的到达和旧数据的删除使滑动窗口内的数据不断更新而导致滑动窗口不再满足隐私要求时,本文分析了删除子窗口和添加子窗口对滑动窗口的影响,同时给出了信息损失度量方法。其次,根据删除子窗口和添加子窗口建立受影响的关联规则树,从而快速找出造成当前窗口不满足ρ-不确定性隐私要求的敏感关联规则,利用抑制方法选择尽可能少的项目进行删除使得当前窗口满足ρ-不确定性。为了减少数据的信息损失,本文进一步提出了抑制和概化相结合的方法,根据信息损失度量来判断选择将项目删除还是概化使得当前窗口达到隐私要求。最后,给出了系统的设计方案以及各模块的详细实现过程。从数据匿名的效率和数据的效用性两个方面,对本文提出的方法与直接利用静态事务型数据匿名方法进行比较,实验结果表明,本文的方法能快速匿名,同时能有效地保证数据的效用性。
[Abstract]:In recent years, with the rapid development of network information technology and its application in people's lives, the Internet has produced a lot of data, which contains a lot of information, business organizations, Government departments can use the data for business decisions, scientific research, and so on. Data sharing is therefore necessary, but the data contains a large amount of personal sensitive information, and if the data owner does not properly handle the data before releasing it, it may cause the privacy information of the individual to be leaked. Thus bring harm to the individual. Therefore, privacy protection in data publishing has become the focus of current research. This problem studies how to process the data so that the published data does not reveal the privacy of individuals and has high availability. At present, privacy protection techniques in data publishing are mostly based on static data sets. With the promotion of the Internet of things and the arrival of the big data era, the network click data, telephone call records, shopping data of large supermarkets and other data are usually shown as dynamic data flow, data stream has mass, real-time and dynamic variability. Traditional privacy protection techniques are not suitable for data flow environments. Transactional data flow is a common form of data flow, such as shopping data in shopping malls, in which each record is a collection of items. This data usually contains sensitive information about the user. At present, privacy protection in data stream publishing is mainly aimed at relational data stream. Because transactional data streams have the characteristics of high and sparse, the privacy protection techniques used to deal with relational data streams can not be directly applied to transactional data streams, and transactional data streams have real-time and dynamic variability. The traditional privacy protection method for static transaction data can not be directly applied to data flow environment. In this paper, the privacy protection of transactional data stream publishing is studied, and a method of privacy protection for transactional data stream publishing based on sliding window is proposed. The main research work is as follows: firstly, combining sliding window and 蟻 uncertainty privacy protection model, a 蟻-uncertainty model based on sliding window is proposed, that is, every sliding window is required to satisfy 蟻-uncertainty. When the arrival of new data and the deletion of old data make the data in the sliding window update continuously, which leads to the sliding window no longer meet the privacy requirements, this paper analyzes the effect of deleting and adding the child window on the sliding window. At the same time, the measurement method of information loss is given. Secondly, the affected association rules tree is established based on deleting and adding sub-windows, and the sensitive association rules which cause the current window not to meet the 蟻 -uncertainty privacy requirements are quickly found. The suppression method is used to select as few items as possible to delete so that the current window satisfies 蟻-uncertainty. In order to reduce the information loss of data, this paper further proposes a combination of suppression and generalizability, according to the information loss measure to determine whether to delete or generalize the items to make the current window meet the privacy requirements. Finally, the design scheme of the system and the detailed implementation process of each module are given. From the efficiency of data anonymity and the utility of data, this paper compares the proposed method with the method of using static transaction data anonymity directly. The experimental results show that the method proposed in this paper can be anonymous quickly. At the same time, the utility of the data can be effectively guaranteed.
【学位授予单位】：广西师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.13;TP309

【相似文献】