Privacy-preserving anonymization algorithm with information loss minimization

对于包含个人信息的数据库表的匿名化，我们提出了一个新的框架，该框架将隐私约束作为个人识别属性和敏感属性之间的推断概率（准确性）给出。同时，将匿名化导致的信息损失建模为成本函数，并提出了一种寻找最优匿名表的算法。作为相关研究，我们还开发了全球分布式环境的访问控制方法，以及社交网络服务中隐私设置的用户协助。

For anonymizing database tables containing personal information, we proposed a new framework such that privacy constraints are given as inference probabilities(accuracies) between person-identifying attributes and sensitive attributes. Also, information loss due to anonymization is modeled as cost functions, and an algorithm that finds an optimal anonymized table was developed. As related research, we also developed access control methods for globally distributed environments, as well as user assistance for privacy settings in social networking services.