Classification and rule updating based on rough set theory in dynamic databases

With the changes of the objective world, new members and new patterns will be appended to databases. Therefore, the former rules sets obtained from the database require updating. In this paper, based on rough set theory, attribute-voting method for rule updating is presented. And, a general strategy for classification and rule updating using rough set theory is put forward. Decision-makers and analysts must distinguish new patterns from others or sort new members into the previous patterns. Therefore, the importance of their subjective initiative is emphasized, instead of depending only on machine. In the end, two examples illustrate that the view is reasonable.