Attribute reduction is an important pre-processing step for data mining, and has become a hot research topic in machine learning, which involves high-dimensional descriptions of input features (attributes). It is therefore not surprising that a lot of research has been done on dimension reduction.

Information systems are the most known forms of knowledge representation, in this paper, we propose a novel attributes' reduction technique for information systems. The proposed technique [namely, morphological accuracy reduction (for short, MAR)] is based on computing a morphological accuracy using the morphological operators (neighborhood-erosion, neighborhood-dilation and neighborhood structure elements). Comparing with reduction using nano topology, experimental results show that the proposed MAR method is an efficient algorithm for attributes reduction and calculating the core attributes. The main advantage of the new method is that it helps to reduce data without losing useful information, as well as it saves time and reaches the best core in fewer steps.