The storage, management, and application of massive spatio-temporal data are widely applied in various practical scenarios, including public safety. However, due to the unique spatio-temporal distribution characteristics of re-al-world data, most existing methods have limitations in terms of the spatio-temporal proximity of data and load balancing in distributed storage. There-fore, this paper proposes an efficient partitioning method of large-scale public safety spatio-temporal data based on information loss constraints (IFL-LSTP). The IFL-LSTP model specifically targets large-scale spatio-temporal point da-ta by combining the spatio-temporal partitioning module (STPM) with the graph partitioning module (GPM). This approach can significantly reduce the scale of data while maintaining the model's accuracy, in order to improve the partitioning efficiency. It can also ensure the load balancing of distributed storage while maintaining spatio-temporal proximity of the data partitioning results. This method provides a new solution for distributed storage of mas-sive spatio-temporal data. The experimental results on multiple real-world da-tasets demonstrate the effectiveness and superiority of IFL-LSTP.

翻译：海量时空数据的存储、管理及应用广泛服务于包括公共安全在内的各类实际场景。然而，由于真实数据独特的时空分布特征，现有多数方法在数据时空邻近性与分布式存储负载均衡方面存在局限性。为此，本文提出一种基于信息损失约束的大规模公共安全时空数据高效划分方法（IFL-LSTP）。IFL-LSTP模型专门针对大规模时空点数据，通过结合时空划分模块（STPM）与图划分模块（GPM），能在维持模型精度的同时显著降低数据规模，从而提升划分效率；此外，该方法还能在保持数据划分结果时空邻近性的同时，确保分布式存储的负载均衡。该方案为海量时空数据的分布式存储提供了新思路。在多个真实数据集上的实验结果表明了IFL-LSTP的有效性与优越性。