【摘要】 图模存储数据之间存在复杂的耦合性,在增量数据差分更新过程中难以准确识别和分离出真正需要更新的部分。当某一数据发生变动时,会影响到多个相关的数据,导致更新过程中出现连锁反应,增加了数据更新的复杂性。为此,提出云-端协同架构下图模存储增量数据的差分更新方法。利用融合图注意力网络与门控循环单元,分别检测图形结构与模型数据变化。运用轨迹系数对数据变化过程中产生的增量数据实施分割处理并采取差分运算,得到增量数据实时状态。组建基于云-端协同架构,在云平台层融入边缘匹配度评估模型,保证云端数据更新完整性,将云端更新后的数据传输至终端设备层,在多个耦合关联数据提出更新需求时,采用云端统一规划和协调更新操作,避免因局部更新导致的连锁反应失控,并利用差分函数完成图模存储数据差分更新操作。实验结果表明,所提方法能精确提取图模数据变化特征,减少数据传输比例的同时提高更新速度,能迅速完成图模存储增量数据更新操作。更多还原

【Abstract】 Complex coupling exists between data stored in graph models,making it difficult to accurately identify and isolate the parts that truly need to be updated during differential updates of incremental data. When a single piece of data changes,it affects multiple related data,leading to a chain reaction during the update process and increasing the complexity of data updates. To this end,a differential update method for incremental data stored in graph models under a cloud-end collaborative architecture is proposed. A fused graph attention network and gated recurrent unit are used to detect changes in graph structure and model data,respectively. Trajectory coefficients are used to segment the incremental data generated during the data change process and perform differential operations to obtain the real-time status of the incremental data. A cloud-end collaborative architecture is established,integrating an edge matching evaluation model into the cloud platform layer to ensure the integrity of cloud data updates. Updated data from the cloud is transmitted to the terminal device layer. When multiple coupled and related data require updates,unified cloud-based planning and coordination of update operations are implemented to avoid uncontrolled chain reactions caused by local updates. The differential update operation of the graph model storage data is completed using a differential function. Experimental results demonstrate that the proposed method can accurately extract the characteristics of graph model data changes,reduce the data transmission ratio,improve the update speed,and quickly complete the incremental data update operation of the graph model storage.更多还原