Generalising the consistent database sampling process

Database sampling is widely used in many database ap-plications when, for efficiency reasons, an entire database cannot be used. This paper analyses the use of consistent sampling, that is, sampling according to certain criteria (e.g. integrity constraints) used to evaluate the consis-tency of the resulting sample. This alternative to random sampling, the most common sampling strategy, is particu-larly appropriate in the context of constructing prototype databases to support Information System Development. The paper firstly presents a framework for evaluation of prototype database construction methods. Then a general description of the Consistent Database Sampling Process is introduced. Finally the paper outlines a sampling tool which implements this Process