Efficient processing of top-k join queries by attribute domain refinement

Abstract

In this paper, we propose a new method for efficient processing of a top-k join query by its translation into a sequence of range queries, which are generated by performing iterative domain refinement of attributes included in the scoring function. In this process, we exploit the statistics for data distributions of the individual attributes, which in the form of histograms are available to an RDBMS. To improve the performance of our method, we use heuristic techniques to minimize the execution cost of range queries and the number of iterations. We use the PostgreSQL query engine optimizer to prove our theoretical results. We have done exhaustive set of experiments by exploiting different input parameters and by using cross checks to prove the results. We have applied our experiments to the TPC-H benchmark data sets, and the results we obtained confirm the efficiency of our approach. © 2012 Springer-Verlag.