大数据蕴含着巨大价值，被称作“未来社会发展的新石油”。数据收集是获取大数据的重要手段。然而，所要收集的数据中往往涉及个人敏感信息，收集这些数据可能会导致严重的个人隐私泄露问题。本地差分隐私是一种专门为解决数据收集导致的个人隐私泄露问题而提出的隐私保护技术，并已受到工业界和学术界的广泛关注。现有的满足本地差分隐私的数据收集工作主要集中在低维数据收集，而对于高维数据收集的研究尚处于起步阶段。目前所提出的满足本地差分隐私的高维数据收集方法要么只能用于收集原始高维数据各维度的特定信息，要么在所收集到的高维数据的效用方面存在严重不足。为此，本项目拟对满足本地差分隐私的高维数据收集问题展开系统、深入的研究，提出解决该问题的新思路和新方法。研究成果将完善满足本地差分隐私的数据收集的理论体系，为满足本地差分隐私的高维数据收集在实际中的应用奠定基础。

Big data contain tremendous value, which are called “new oil for future social development”. Collecting data is an important way to get big data. However, since the data to be collected often involve sensitive personal information, collecting such data may lead to serious problems of privacy leakage. Local differential privacy is a privacy protection technology specifically designed for solving the problem of privacy leakage caused by data collection, and has attracted much attention from both the industry and academia. Existing studies on locally differentially private data collection mainly focus on low-dimensional data collection, while the research on high-dimensional data collection is still in its infancy. The proposed methods for locally differentially private high-dimensional data collection either can only be used for collecting the specific information of each dimension of the original data or have serious deficiencies in the utility of the collected high-dimensional data. To this end, this project intends to conduct a systematic and in-depth study of the locally differentially private high-dimensional data collection problem, and proposes new ideas and new methods for solving this problem. The research findings of this project will enrich the theoretical system of locally differentially private data collection, and build a solid foundation for the usage of locally differentially private high-dimensional data collection in real-world applications.