导　　师： 刘洪伟

学科专业： 1201

授予学位： 硕士

作　　者： ;

机构地区： 广东工业大学

摘　　要： 信息技术的不断进步推动着信息化下的协同计算不断地向前发展,参与者通过在网络环境中“交流”信息可以完成一些复杂函数的计算。但是,各个参与计算的参与者所拥有的数据一般属于个人私有的信息或者是涉密信息,因此隐私保护问题成为了协同计算发展所面临的重要的制约因素。让参与者“贡献”自己的私有数据参与协同计算并且保护其私有信息不泄露给其他参与者已经成为众多研究者的研究目标,也取得了一定的成果。1982年,A.C.Yao最先给出了安全多方计算的概念,目的就是保护私有信息的协同计算问题,这一理论及方法引起了众多研究者的高度关注,并继承和发展了大量的安全多方计算理论及其应用方面的研究。 安全多方计算是现代密码学研究的一个重要分支,是信息安全中的一个重要的研究领域。其主要思想是：在一个分布式网络里,两方或多方根据他们的秘密输入而执行一个算法,使得在各方都能够得到正确输出的同时又可以保护自己输入信息的隐私性。 协同过滤已经成为了个性化推荐系统的一种主要工具。然而协同过滤中存在的数据稀疏问题严重影响着推荐的效果。目前越来越多的电子商务网站在寻求同类型网站之间的合作,尝试进行跨网站的协同过滤。 本文针对上述问题所做的工作主要有以下几点： 1/)论述推荐算法的发展现状、经典的推荐算法以及常用的隐私保护算法； 2/)在传统的用户评分的基础上对协同过滤推荐算法进行了改进； 3/)重点讨论安全多方计算协议,列举安全多方计算的相关基础协议及其应用,设计出一套基于安全多方计算的协同过滤推荐算法,并进行了安全性及正确性分析； 4/)设计了一个安全多方协议解决了商业决策中的销售总量的问题,并进行了实例验证,且对方案的正确性、安全性和复杂度进行了分析。 The progress of information technology to promote the informationization of collaborative computing constantly moving forward, Participants can do some calculation of complex function by the "communication" in the network environment information.However, all involved in the calculation of the participants generally belong to personal private information or data, the problem of privacy protection becomes an important restriction factor for the development of collaborative computing.The participants contribute their own private data,and protect their private information is not leaked to other participants which has become the research goal for many researchers.In1982, A.C.Yao gave the concept of secure multi-party computation first,the theory and method have attracted many researchers' attention,the inheritance and development of a large amount of secure multi-party computation theory and application research. Secure multi-party computation is an important branch of modern cryptography research and research field in information security.The main idea is following:in a distributed network, two or more parties carry out an algorithm that makes it possible to get the correct output in the parties according to their own secret input and implement at the same time it can protect their input information of privacy. Collaborative filtering has become a major tool of personalized recommendation system.However, the sparseness problem of data in collaborative filtering has affected the effect of the recommendation.Currently, more and more e-commerce sites looking for the same type of cooperation between sites. Based on the above problems,this paper's work mainly include several aspects as following: 1/) Discourses the current development situation of the recommendation algorithm, the classic recommendation algorithms and the algorithms for privacy protection; 2/) On the basis of the traditional user ratings for collaborative filtering recommendation algorithm makes some improving; 3/) Focus on secure multi-party computation protocol, lists basic protocol and application of secure multi-party computation, designs a set of collaborative filtering recommendation algorithm based on secure multi-party computation and analyzes the security and validity; 4/) Design a secure protocol to solve the problem of the total sales in business decisions, make a example verification,and analysis the correctness of the scheme, the security and complexity.

关 键 词： 安全多方 隐私保护 协同计算 推荐算法 协同过滤

分 类 号： [TP309 F724.6]

领　　域： [自动化与计算机技术] [自动化与计算机技术] [经济管理]