多数据源上Top-k中间模式的产生算法

Generating Top-k Mediated Schemas for Multiple Data Sources

模式集成在很多数据库相关领域起着关键作用,例如数据空间、数据仓库和电子商务等.提出一种自动的多个中间模式的产生方法.首先,引入概念图在抽象层次上表示待集成的多个源模式.其次,给出一种概念之间相似性的划分方法,每种划分方式表示一种源模式的集成策略.最后,利用模拟退火算法在候选中间模式空间中进行搜索,该算法能够自动地找到k个最好的候选中间模式.实验表明,提出的算法是有效的,并且具有较小的运行开销.

Schema integration is a critical step in many database applications,such as dataspace,data warehousing and electronic commerce,etc.This paper proposed an automatic approach to generate the mediated schemas over a set of source schemas.Firstly,the concept graph is presented to represent the source schemas for the unified representation.Secondly,the similarity between concepts is divided into intervals for the generation of the three merging strategies.Finally,the simulated annealing algorithm is employed to automatically generate the best k mediated schemas.Through extensive experiments,the results show that the algorithm proposed is effective and the running time is little.