Database redundant attribute detection using fractal dimension

The method for detecting redundant attributes in relational datasets using the fractal ideology is studied. Based on the fractal dimension of a dataset and its variations, an algorithm for detecting redundant attributes is presented. The work has the following features: datasets with numeric and discrete attributes can be processed; an approach based on depth-equal data dimension division(i.e., the number of attribute values in each interval of a divided dimension is the same) is introduced for computing approximate fractal dimension of a dataset; and the existed algorithm for detecting redundant attributes is expanded, which can discover correlated attribute pairs in a dataset. The experimental results prove the validity of the proposed algorithms.