Discovering non-constant Conditional Functional Dependencies with Built-in Predicates

In the context of the data quality research area, Conditional Functional Dependencies with built-in predicates (CFD p s) have been recently defined as extensions of Conditional Functional Dependencies with the addition, in the patterns of their data values, of the comparison operators. CFD p s can be used to impose constraints on data; they can also represent relationships among data, and therefore they can be mined from datasets. In the present work, after having introduced the distinction between constant and non-constant CFD p s, we describe an algorithm to discover non-constant CFD p s from datasets.