论文部分内容阅读
针对现有并行挖掘技术需要建立在专门的并行机上而使中小图书馆难以实现的弊端,通过研究关联规则的内在并行性,提出一种基于图形处理单元技术的快速并行关联规则算法,实现在单处理器上并行数据挖掘。通过仿真实验证明,优化后的算法在不增加硬件设备的前提下实现单处理机并行,提高运行速度,解决传统并行关联规则挖掘难以应用化的问题,该算法对于数字图书馆具有重要的理论和应用价值。
Aiming at the drawback that the existing parallel mining technology needs to be built on a special parallel machine to make the small and medium libraries difficult to be realized, this paper proposes a fast and parallel association rules algorithm based on the graphics processing unit technology by studying the inherent parallelism of association rules, Parallel data mining on processors. The simulation results show that the optimized algorithm achieves single processor parallelism and increases the running speed under the premise of not adding hardware devices, and solves the problem that traditional parallel association rule mining is difficult to be applied. The algorithm has important theoretical and practical significance for digital library Value.