A study of comprehensive evaluation method based on neighborhood covering information system

In order to deal with the correlation of indexes in comprehensive evaluation problems, we propose a method to measure the importance of the attributes sets based on numerical-valued decision information system. In this method, neighborhood is employed to describe the similarity between the elements, and the positive region in covering rough sets is used as knowledge carrier of the decision information system. Then the importance metric is established according to the changes of knowledge carrier. Furthermore, we construct a comprehensive evaluation model by taking Choquet fuzzy integral as an information fusion tool. Finally, an illustrated example demonstrates the characteristics and effectiveness of the proposed method. The results show that this evaluation method has essential differences from the existing ones and also has good operability and interpretability. Moreover, it can be applied in the fields such as resource allocation, performance assessment and complex system optimization.