Optimizing N-Degree Union Problems in Graph Data Processing for Enhancing the Efficiency of Dynamic Graphs

Graph data structures have found significant applications across various applications, particularly in constructing accurate recommendation systems on social media platforms, such as connection recommendations on LinkedIn and friend recommendations on Facebook. However, we observe that intersecting two imbalanced relationships, referred to as the $\mathbf{N}$-degree union problem, leads to exponential increases in data volume and performance bottlenecks as the degree of relationships grows. Hence, we propose a data pre-processing mechanism to rebalance these imbalance relationships. The rebalanced relationships significantly decrease the memory space demand and alleviate performance bottlenecks in memory-constrained environments. The experiments show that our strategy effectively eases the requirement of memory space and mitigates the performance bottleneck in memory-constrained settings.