Attribute reduction, which is widely used in data mining, is one of the most important issues in the rough set theory. In this paper, the authors present an overview of extended rough set theory which is based on the combination of rough set and the fuzzy set theory to illustrate the opacity and uncertainty of data. In particular, in the first section, the authors present an overview of rough set and fuzzy set theory, in the next section, the authors remind an attribute reduction algorithm which have been proposed on the measures of the dependence of attribute sets in the fuzzy rough set. Accordingly, the authors propose an extent algorithm which is based on the caculation of dependence of attribute sets on each attribute. the algorithm is more efficiency in some cases where the decision class labels are difficult to determine or insufficient.