Algorithms for database keys discovery assistance

Abstract

Only humans may decide whether a column or a set of columns of a table should store minimally unique values. Not adding to a table any existing constraint (business rule), which includes keys, allows for storing implausible instances in it. Unfortunately, 2n possibilities should be considered in the worst case for a table with n columns in order to discover all of its keys. This paper presents and discusses contrastively three as efficient as possible, sound, and complete algorithms that assist database designers in discovering all existing keys in the corresponding subuniverse of discourse.