Applying Bayesian networks for query selectivity estimation in query optimizer of Oracle DBMS
Dariusz Rafał Augustyn, Łukasz Warchał · 2011
The paper presents applying Bayesian network-based method of a selectivity estimation. The query selectivity allows estimate query result size, which allows to choose the optimal method of query execution. Obtaining the selectivity for a query with a selection condition based on many attributes, requires an estimator of a multidimensional probability density function of attribute values. Bayesian network can be used as a memory-efficient representation of the multidimensional distribution of attribute values. The article shows Bayesian network approach applied for extending the functionality of the query optimizer. Some Weka modules are used for implementing Bayesian network-based selectivity estimation in Oracle DBMS optimizer.