Graph pattern matching is a fundamental problem in graph mining tasks. GraphMini introduces proactive pruning using auxiliary graphs to optimize set operations, achieving remarkable speedups compared to existing systems like Dryadic and GraphPi. The system's innovative approach involves online pruning during query execution, leading to substantial performance improvements.
The paper discusses the challenges of graph pattern matching and the opportunities presented by the GraphMini system. It explains the concept of auxiliary graphs and how they are used to accelerate set operations by reducing adjacency list sizes. The authors propose a cost model to estimate the benefits of pruning adjacency lists and introduce compile-time optimizations such as nested parallelism for workload balancing.
The evaluation section compares GraphMini with Dryadic and GraphPi on real-world data graphs, showcasing superior performance in both vertex-induced and edge-induced pattern matching scenarios. The results demonstrate that GraphMini achieves significant speedups, especially in edge-induced pattern matching, highlighting its effectiveness in optimizing graph pattern matching tasks.
Para outro idioma
do conteúdo fonte
arxiv.org
Principais Insights Extraídos De
by Juelin Liu,S... às arxiv.org 03-05-2024
https://arxiv.org/pdf/2403.01050.pdfPerguntas Mais Profundas