Efficient computation of subqueries in complex OLAP

Expressing complex OLAP queries involving nested expressions using normal group-by, aggregation, and joins can be extremely difficult. We propose a technique that translates nested query expressions into an algebra extended with a complex OLAP operator. The GMDJ is an operator with a simple and easy to optimize implementation that is particularly useful for OLAP computations because the size of intermediate results is bound by the size of the base-value argument relation. We show that all SQL subqueries can be expressed in the algebra using GMDJs. This not only makes it easy to integrate subqueries into any query engine that supports GMDJs, but also gives access to a broad range of OLAP optimization strategies for evaluating subqueries. We discuss the coalescing of GMDJs and the completion of tuples, two GMDJ optimizations that are particularly relevant to subquery processing. Our experimental results demonstrate the validity and efficiency of our approach for computing subquery expressions.