拓扑关系查询中决策树生成算法的改进  

Improvement of decision tree building algorithm in topological relations query

作　　者：王宏艳[1] 

机构地区：[1]燕山大学经济管理学院,河北秦皇岛066004

出　　处：《计算机工程与设计》2011年第4期1334-1337,共4页Computer Engineering and Design

摘　　要：为了高效的查询空间数据库系统中对象之间的拓扑关系,摒弃传统决策树生成算法中对空间中各个拓扑关系具有相同出现率的理想化假定,根据实际情况为空间数据库中的拓扑关系的出现率赋值,提出了基于9交集最小子集的决策树生成算法,保证出现率最高的拓扑关系在查询时具有最少的计算次数,使得决策树有更小的带权路径,决策树更接近最优,从而提高了拓扑关系查询效率。实验结果表明,该算法在查询效率上优于传统的决策树生成算法。In order to improve the query efficiency of topological relation between objects in spatial database system,the idealized as-sumption of the same occurrence rate for the topological relations in the traditional decision tree building algorithm is abandoned.The occurrence rates of the topological relations in spatial database will be assigned values in terms of the actual situation.The decision tree building algorithm based on minimal set of 9-intersection model is put forward,which assures that the topological relation with the highest occurrence rate has the least calculation times during the process of the query.As a result,the decision tree built by the new algorithm has a lower weighted path length（WPL）,and closer to optimal decision tree.Therefore,the query efficiency of the topological relations is improved.The experimental result shows that the new proposed algorithm outperforms the traditional decision tree building algorithm.

关 键 词：空间数据库 拓扑关系查询 决策树 9交集最小子集 出现率 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]