A Multi-copy Join Optimization of Information Integration Systems Based on a Genetic Algorithm
A Multi-copy Join Optimization of Information Integration Systems Based on a Genetic Algorithm
复制标题
基于遗传算法的信息集成系统多副本连接优化
DOI：
--
复制
发表时间：
2008
期刊：
影响因子：
--
通讯作者：
Jianzhuo Yan
中科院分区：
文献类型：
--
作者：
Liying Fang;Pu Wang;Jianzhuo Yan
针对异构信息集成系统中局部数据源不可避免的冗余问题，提出了一种基于遗传算法的多副本连接优化方法（MuCoJo）. MuCoJo可以选择合适的冗余副本参与联合查询，并优化联合查询的连接顺序，通过使用冗余副本，MuCoJo扩大了搜索空间，使得不同本地源的并发执行得到最佳利用。同时，MuCoJo可以利用这些系统中的冗余特性，获得更快的联合查询响应时间。实验结果表明了MuCoJo的计算效率及其在信息集成系统中的必要性。此外，在种群初始化过程中，控制无效解可以有效地减少搜索空间，但初始化时间消耗较大。
In view of inevitable redundancies in local data sources in heterogeneous information integration systems, a multi-copy join optimization method (MuCoJo for short) based on a genetic algorithm is proposed. MuCoJo can choose appropriate redundant copies of the tables to participate a joint query and optimizes the join order of it. By using the redundant copies, MuCoJo enlarges the search space so that the concurrent executions of different local sources can be best used. Meanwhile, MuCoJo could take advantage of redundancies features in such systems and get faster joint query response time. Experimental results show the computational efficiency of the MuCoJo and its necessity in information integration system. Moreover, during the population initialization, controlling invalid solutions can reduce the search space effectively at the cost of initialization time consuming.