网格数据检索中结果集的合并算法

Efficient Query Processing in Grid Data Retrieval System

随着网格从科学计算转到企业级应用,要求数据库提供多种服务支持以实现更强更丰富的资源共享和应用。网格上的数据库只能通过网格服务进行访问,而数据库中的数据也只能通过网格服务接口来存取。因此如何在网格环境下直接对分布在各地的数据库进行高效的检索就是迫切要解决的问题。本文首先提出了一个网格环境下数据检索的体系结构,然后针对该结构下的数值型数据的Top-k查询问题给出了 GrangM算法,它有效解决了来自不同数据源查询结果的合并问题。对该算法的模拟实现表明,它可以快速、高效地合并网格中多结点检索出的结果,减少连接中间结果的大小,降低发送查询请求的通信量。

In many Grid applications,the arrtributes for which users specify target values might be handled by exter- nal, auntonomous Grid peer sources with self-interfaces. This paper tries to introduce Top-k query in Grid Data Re- trieval System. First, a distributed query processing model for Grid is proposed. Ranking and merging of results are distributed across the Grid Service peers. Next, an efficient Top-k query results merging algorithm (GrangM) is pro- posed. By the fore-statistical information, the most possible sources are selected to execute the query, so as to greatly improve the search efficiency. Experiments showed that such Top-k query improves the query effectiveness and effi- ciency.