QCUR树

找到一个高效的索引结构一直是空间数据库研究的重点。CUR树突破传统思维,提出用一个代价函数来决定叶子的高度,从而在整体上优化了索引结构的性能。然而,它基于的查询分布模型是静态的,这限制了它的应用和发展。文章提出的QCUR树在CUR树的基础上,采用了半动态式的查询分布模型。让QCUR树可以根据查询分布的变化调整代价函数,从而可以根据查询分布的变化来优化树的性能。实验也说明,在动态的查询分布下,QCUR树性能优于CUR树。

Query-directed Cost-based Unbalanced Tree

It is always the key technology of spatial DBMS to find a efficient index structure.By breaking the tradition-al law,CUR-trees use one cost function to guide the height of the leaves.A CUR-tree is built taking into account a given query distribution for the queries and a cost model for their execution.Depending on this cost function,CUR-trees can optimize the performance of the searching.However,the query distribution model on which is depended by CUR-trees is a static model.So it's less useful under active conditions.This paper presents a new index structure,QCUR-tree,which use a semi-active query distribution model.So QCUR-trees can modify the cost function according to the query distribution fluctuate.And our experiments show that under active conditions,QCUR-trees can provide more efficiency than CUR-trees.