Study On Efficiency Of Method For Computing RS Core Based On Program Transformation

Rough Set a new technology for Data Mining,which analyzing and processing incomplete, inconsistent and inaccurate data then reducing knowledge without prior knowledge. Attribute reduction, the key step of which is seeking attribute cores, is major research contents of Rough Set. The most of reducing algorithms begin with attribute cores though heuristic search to the minimum reduction. Seeking attribute cores based on differential matrix is common method, which is improved by many scholars but has low efficiency.Program transformation, as a new direction and way of improving the efficiency of algorithm has always caused the extensive concern at home and abroad. Using program transformation to improve the efficiency of seeking attribute cores algorithm based on differential matrix is a new direction.Centering on strategies of program transformation to improve the efficiency of algorithm in this paper, using that to improve the efficiency of seeking attribute cores algorithm based on differential matrix.The main studies of this paper are:(1) Principle of seeking attribute cores algorithm based on differential matrix is described, the latest research on seeking attribute cores improved algorithms based on differential matrix are analyzed. The reasons of low efficiency of algorithms are summarized, which offer idears for reducing the complexity of algorithm.(2)According to characteristics of application methods of program transformation, specifily describes seeking attribute cores algorithms based on differential matrix by Haskell functional language.The efficiency of algorithm could not considered at first,but correctness of which is guaranteed, that is basic of transformation.(3)After analyzing and processing the specifily describes seeking attribute cores algorithms based on differential matrix, combined with the reasons of low efficiency of algorithm,using Tupling and Accumulation strategies of program transformation to improve the efficiency of algorithm.(4)After analyzing the improved algorithm by strategies of program transformation, using a method to divide data set into subsets to improve the efficiency of creating differential matrix, then combines with Accumulation strategies to improve the algorithm.The improved algorithm after program transformation compares with original seeking attribute cores algorithms based on differential matrix, the former has higher efficiency. Computer simulation shows that the correctness and high efficiency of algorithm proposed in this paper.