面向时空特性的社交网络推演攻击与隐私保护关键技术研究

项目名称： 面向时空特性的社交网络推演攻击与隐私保护关键技术研究

项目编号： No.61502316

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 自动化技术、计算机技术

项目作者： 刘向宇

作者单位： 沈阳航空航天大学

项目金额： 20万元

中文摘要： 推演攻击是一种基于概率模型、数据推演等技术的隐私获取方法，而签到数据的时空特性为推演攻击提供了更多的隐私攻击方法和更强的隐私攻击能力，使得现有技术在防止推演攻击导致的社交网络隐私泄露方面存在不足。课题从位置隐私、模式隐私、关系隐私及个性化数据发布等四个方面，对面向时空特性的社交网络推演攻击与隐私保护关键技术进行研究。研究时空约束条件下的隐式位置隐私保护技术，实现通用隐式位置隐私威胁评估与保护；提出自适应时空模式隐私保护技术，支持时空模式隐私泄露实时检测与保护；研究面向时空关联性的敏感关系隐私保护技术，实现基于时空关联性特征设计链式敏感关系推演及保护机制；研究支持个性化发布约束的签到数据隐私保护，实现隐私保护强度与数据共享精度的动态调控。课题最终实现一个面向时空特性的社交网络数据安全发布原型系统，研究成果将在理论和实践两个层面，为新型社交网络的数据隐私保护与安全共享提供良好支持。

中文关键词： 社交网络；推演攻击；隐私保护；时空特性

英文摘要： Inference attack is a privacy achieving method based on probabilistic model and data inference technologies. Due to the spatial-temporal features of check-in data in social networks, inference attacks are provided with more methods and power on achieving privacy information. The present privacy protecting methods could not prevent the privacy leakage due to the inference attacks based on spatial-temporal features. This research proposal proposes studying on spatial-temporal features oriented key techniques of inference attacks and protecting privacy in social networks from the aspects of location privacy, pattern privacy, relationship privacy and personalized data publication. We propose the protection of hidden location privacy under the spatial-temporal constraints, which builds a general framework of evaluating the leakage risk and providing protection. We study the adaptive protection of spatial-temporal pattern privacy, which supports the real-time detection of privacy leakage. We also propose the protection of sensitive relationships based on spatial-temporal correlations, including the design of chain mechanism on sensitive relationship inference and protecting. We study the personalized check-in data publication, which dynamically adjust between the privacy protecting strength and data share precision. We plan to implement a spatial-temporal features oriented prototype system of secure publishing social network data. Our research results would provide theoretical and practical supports for privacy protection and secure data sharing in new type social networks.

英文关键词： social network;inference attack;privacy protection;spatial-temporal feature