| Rough set theory is initialized by Professor Pawlak in 1982. It is a new mathematical tool to deal with problems on vagueness and uncertainty. The research of theory and applications has become a hot spot at home and abroad. The application of rough sets theory for knowledge discovery, data reduction, decision support, classification, pattern recognition and others have proved to be a very effective new mathematical approach.In decision support systems and other intelligent systems, with the knowledge acquired growing rapidly, the amount of data becomes larger and larger. But not all the data are useful for our decision making. How to reduce the data that are useless and not affect the decision ability is the critical problem to be resolved. One of the most effective way to do that is using rough set theory.Integrated the theory and the fact, our paper has done some research deeply on how to use Rough set theory to reduce the data in decision support systems. In Rough set theory, there is no unified definition on attribute significance. Our paper proposes to use the function f_B(a) as the metric of the degree property importance degree, and gives an approach on reducing the data based on the function. The actual examples have proved it practicable. With the problem that the decision table in practice are always inconsistent, there is no a good way to deal with up to now. Our paper proposes the conception of loose dependence degree, which lowers the severe requirement of definition on dependence degree in classic rough set theory, and gives the conception of condition property redundancy and reduct based on loose dependence degree, and also gives the reduct algorithm used to make the property set reduced. Based on theory research and actual examples analysis, we get the result that the reduct algorithm based on loose dependence degree has better fault tolerance ability and knowledge discovery independently ability. It improves the decision information system's adaptive ability. |
PDF Full Text Request