Author(s)

Liu Bohong, Jiang Xinyuan

Corresponding Author

Liu Bohong

Abstract

DFD is a depth-traversal functional dependencies discovery method, it does not consider association between nodes of power set lattice. We improved DFD by using attribute information entropy combined with DFD to reduce the repeated frequencies of traversals. Datasets of UCI are used to verify that the improved DFD runs faster than original DFD.

Keywords

Functional Dependencies, Attribute Partition, Information Entropy.