面向个性化推荐服务用户隐私保护的若干关键问题研究

Personalized recommendation has demonstrated its effectiveness in improving the problem of information overload on the Internet, and has achieved great success in many application fields (typically, e-commerce). However, to achieve accurate recommendation, it needs to collect a great deal of users’ private information, resulting in the problem of personal privacy and security, which has become a major barrier for the development and application of personalized recommendation. In this project, by using semantic computing, information confusion, index encryption etc., we will study the corresponding protection approaches for main kinds of channels of user privacy leaks in a personalized recommendation system, so as to ensure the security of user privacy on the untrusted server-side, under the precondition of not compromising the accuracy and practicality of personalized recommendation. To this end, in this project, under a client-based architecture, we will study: (1) a cover-up approach for user sensitive preference protection, so as to ensure the security of sensitive preferences and the accuracy of personalized recommendation; (2) a confusion approach for user query action protection, so as to prevent the untrusted server-side from continuing to record and analyze user query intention; and (3) an encryption approach for user basic data protection, so as to ensure the security of user data and the effectiveness of data queries. Finally, by means of the implementations of these approaches, we aim to disrupt the main channels of user privacy leaks in a personalized recommendation system, so as to lay the foundation of building a secure and effective personalized recommendation system.

个性化推荐是解决互联网时代信息超载问题最有效的工具之一，已被广泛地应用于电子商务等领域。然而，为了获得准确的推荐，它需要收集大量的用户个人信息，从而引发用户隐私安全问题，严重地制约了个性化推荐技术的发展与应用。本项目将结合语义计算、信息混淆、索引加密等技术理论，针对个性化推荐系统中各类用户隐私泄露通道，研究相应的保护方法，以期在不损害推荐服务的准确性和实用性的同时，确保各类用户隐私在不可信服务器端的安全性。基于客户端的体系结构，本项目将主要研究：（1）用户敏感偏好掩盖保护方法，以确保敏感偏好的安全性和推荐服务的准确性；（2）用户查询行为混淆保护方法，以阻止服务器端继续收集分析用户查询意图；（3）用户基本属性索引加密方法，以确保用户属性数据的安全性和数据查询的有效性。最后，通过这些隐私保护方法的实施，阻断个性化推荐系统中用户隐私泄露的主要通道，从而为实现安全有效的个性化推荐服务奠定基础。

个性化推荐是解决互联网时代信息超载问题最有效的工具之一，已被广泛地应用于电子商务等领域。然而，为了获得准确的推荐，它需要收集大量的用户个人信息，从而引发用户隐私 安全问题，严重地制约了个性化推荐技术的发展与应用。本项目结合语义计算、信息混淆、 索引加密等技术理论，针对个性化推荐系统中各类用户隐私泄露通道，研究相应的保护方法，以期在不损害推荐服务的准确性和实用性的同时，确保各类用户隐私在不可信服务器端的安全性。基于客户端的体系结构，本项目主要研究了：（1）用户敏感偏好掩盖保护方法，以确保敏感偏好的安全性和推荐服务的准确性；（2）用户查询行为混淆保护方法，以阻止服务器端继续收集分析用户查询意图；（3）用户基本属性索引加密方法，以确保用户属性数据的安全性和数据查询的有效性。最后，通过这些隐私保护方法的实施，阻断个性化推荐系统中用户隐私泄露的主要通道，从而为实现安全有效的个性化推荐服务奠定基础。