Method And Application Of Knowledge Reduction Based On Rough Set

Rough set is a data analysis theory for dealing with incomplete and uncertain knowledge, and knowledge reduction is a significant branch of rough set which has been applied to many areas successfully including knowledge acquisition, decision support and pattern recognition. Research on the method of knowledge reduction based on rough set can provide new clues and methods for incomplete and uncertain problem.Firstly, the comprehensive survey about the common clues of knowledge acquisition by making use of rough set is given. For the method of attribute reduction based on the significance of attribute is imperfect for inconsistent decision table, the disadvantage is indicated that inconsistent records have an effect on correctness of the results. Based on the study above, on basis of the theory that whether the records of generalized characteristic table are deleted is decided by change of positive region, positive region is designed to enhance the fitness of reduction model, and the character of inconsistent record is propitious to exert the virtues of reduction model. Furthermore, improved algorithm is proposed which is suitable for consistent and inconsistent decision table to enhance the generalization ability. Subsequently the capability of algorithm is analyzed and the results indicate that it is more effective. And then the completeness of heuristic algorithm is considered to define importance of value, and importance of value is weighed by change of mutual information between condition and decision attribute, so improved algorithm based on condition entropy is proposed to enhance efficiency and neatness. The quality of decision rules is estimated by attribute evaporation rate and confidence level to show the fitness of improved algorithm.At last, according to the demand of enterprise, lots of data of customers are changed into pattern of characteristic of customers, and decision table is reduced to gain rules, and the function is realized that obtaining implicit knowledge, and then it can provide powerful basis for decision of enterprise and improve customer relationship management.Method of knowledge reduction based on rough set is designed in which whole reduction is combined with part reduction, and objective choice is combined with subjective judgment. It not only improves the disadvantage that wrong results are gained because of inconsistent records, but also solves the problem of high calculation cost. With the practical application, the method can generate effective decision rules and good effects are acquired.