一种适用于点和区间混合型维度数据集的多维索引

A Multidimensional Indexing for Data Sets of Point and Interval Dimensions

点和区间混合型维度数据集是空间数据库系统和GIS中重要的数据对象。在分析研究R*树和SS树的基础上,提出了一种适用于索引点和区间混合型维度数据集的索引结构——PI树。PI树利用超球划分数据集的多维空间,以提高结点存储利用率,从而降低数据插入时的I/O次数。文章给出了PI树插入、删除和查询算法的形式化描述。理论分析和实验结果表明,所提的PI树性能上总体优于R*树。

Data sets of point and interval dimensions are important in spatial database system and GIS. This paper proposes an index structure PI-tree for the data sets of point and interval dimensions, based on an analysis of R^＊ -tree and SS-tree, of which the former is used to demarcade multidimensional space of data set using hyper-sphere with the aim of improving accessing and reducing the I/O times of inserting data. Furthermore, the algorithm of insertion, deletion and retrieval of H-tree is also presented. Finally, the analysis and results of experiment show that PI-tree outperforms R^＊-tree.