.png){kind=icon}
.png){kind=icon}
Return
Dynamic Database Partitioning with Multi-Task Graph Attention and Incremental Updates
.png){kind=icon}
PRE
AI
DOI：10.1016/j.eswa.2026.132706.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• MTGA dynamically integrates UF/QC into column-level graph for real-time partitioning. • Local subgraph refresh avoids full graph reconstruction, reducing update overhead. • Multi-task graph attention jointly optimizes throughput, latency, and load balancing. • Importance sampling improves long-tail query handling via load-aware evaluation. • Event-driven triggers enable on-demand partitioning strategy adjustments.
Keywords:
Dynamic Database Partitioning
Multi-Task Graph Attention
Incremental Updates
Real-Time Partitioning
Load Balancing
Journal
IF:
7.5
Papers:
2.9W
Citations:
10.2W