摘要在当今社会的信息交流中，信息在不同使用者之间传递和在公共平台上的共享越来越常见，人们不再只拘泥在自己拥有的信息范围内，将自己的信息发布出去和从外界获取信息成为了做许多其他行为的基础工作。尽管在发布信息之前，发布者通常会将明显可以标识出某一个对象的属性删除或者以不表示任何信息的符号代替这类属性的值，但在攻击者的各种攻击手段下，信息发布背后仍然有着很大的隐患。1998年，针对这种信息发布的不安全性，P.Samarati和L.Sweeney首次提出了K-匿名模型。在以后的发展中，为了平衡信息的清晰性、数据的可用性和信息的隐私性， K-匿名模型有了不断的改进。64617

本文综合考虑了经过K-匿名过程后信息的可用性和隐私的保护程度，使用分组的方法将差别较小的元组归为一类，将整个数据表的元组分为若干类，K-匿名的行为将在每个组分别进行，有效避免了整体K-匿名带来的较深层次泛化使得原始数据不清晰的问题。本文还将敏感属性按照敏感程度分组，以免过多的同等程度的敏感属性在同一个等价类中，遭受同质攻击时泄露信息的发布者不愿意泄露的敏感信息。

毕业论文关键词：  隐私保护    K-匿名模型    泛化

毕业设计说明书（论文）外文摘要

Title   The Research and Implementation of Privacy in   Data Release                                                  

Abstract

In the exchange of information in today's society, the information dissemination between the different users and information share on a common platform are becoming more and more common, people are no longer only bound  to their own information  , publishing their own information  and get information from the outside world have been a access to to do the other things. Although the publisher usually will delete some obvious identifiers which can identify a object or  use the symbols which do not mean anything to take the place of the value of those attributes before

release the information, there are still a lot of hidden dangers under

attacking. In 1998,P.Samarati and L.Sweeney first proposed K-anonymity model against this information insecurity. In the future development,

K-anonymity model will improve in order to balance the clarity of the information, the availability of data and information privacy.

  In this paper, I consider the availability of information and the degree of protection of privacy after K-anonymity, using the grouping method to classify the similar tuples  as a group, and the entire data table 

is pided into a number of categories, then use the K-anonymity in each class respectively, this kind of operation effectively avoid the situation 

which the original data becomes  unclear after the deeper generalization  which overall K-anonymity makes . This article also pide sensitive attributes according to the degree of sensitivity, in order to avoid too much sensitive properties of the same degree in the same category which probably leak sensitive information that publishers are reluctant to under homogeneous attack.

Keywords：   Privacy Protection    K-anonymity Model    Generalization          

1绪论 1

1.1研究背景及意义1

1.2国内外研究现状1

1.3研究内容3

1.4论文组织结构3

2匿名的隐私保护 4

2.1 数据发布中的攻击 4

2.2 匿名化技术6

2.3 匿名的隐私保护模型  7

2.3.1 K-匿名模型10