改进的基于R树的空间连接代价模型

Improved cost model for spatial joins using R-trees

针对Yannis Theodoridis等人提出的空间连接代价模型存在比较理想化的限制条件——假设数据均匀分布,缓冲策略使用简单的缺点,利用划分子空间并抽样获取非均匀数据实际密度的策略,提出了优先保存查询集合树的最新访问路径的有效中间节点的缓冲区算法,给出了改进后的评估公式。实验结果表明,改进后的模型比原模型提高了评估的精确度。

Thinking on ideal strict conditions about cost model proposed by Yannis Theodoridis, that is the defect of symmetrical data and simple buffer strategy. Using the strategy of partition subspace and acquiring practical density of heterogeneous data by sampling, buffer method of the valid middle node in the latest accessing path of firstly saving inquiring set tree is proposed, improved evaluation equation is put forward. The experimental results show that the improved cost model improves the evaluation precision than the original cost model.