实例化视图选择算法PBC

在数据仓库中,如何选择实例化视图是一个重要的问题。针对一类特定的数据立方体,该文提出了一个基于代价策略的实例化视图选择算法。通过对一个实际数据集的分析,发现在数据立方体中有很多父子视图具有相同的体积,其原因是用于产生数据立方体的基本关系的属性之间存在着密切的联系。显然,对这类数据立方体不能像算法PBS那样按照体积的大小来选择要实例化的视图。为此,设计了算法PBC,不但可以快速地给出满足条件的实例化视图集,而且可以准确地找到具有最短平均响应时间的实例化视图集,避免了在用户给出过大的参数时,实例化一些无益于缩短查询响应时间的视图。实验结果表明,算法PBC是有效的。

PBC:A Materialized Views Selection Algorithm

It is an important issue that how to select views to materialize.This paper proposes a view selection algorithm based on query cost,which is suitable to a certain type of data cube.By analyzing a real data set,authors find that there are lots of views that have as many tuples as their parent.The reason is that there are correlations between attributes of base relation used to produce data cube.Clearly,in this case,people can't use algorithm PBS,because it selects views to materialize based on view size.Authors design an algorithm PBC.It not only can find the materialized views satisfied given condition rapidly,but also can correctly find the materialized views that make response time minimal.It avoids materializing views that are no benefit to reduce response time when users give improper parameter.The results of experiment show that the algorithm is effective.