Research On Algorithm Of Attribute Reduction Based On Rough Set

With the development of science and technology, the information that need to process is increasing rapidly. It is important to find a useful method to obtain valuable information from a great deal of data. Knowledge Discovery is the process through which we can extract hidden but useful knowledge from half-backed and random data. Rough Set is a new mathematical tool to deal with uncertain and fuzzy knowledge. Poland scholar named Pawlak.Z firstly proposed the theory of rough set in 1982. The attribute reduction based on Rough Set is the important content in Knowledge Discovery. It is also a NP-hard question.Some useful attribute reduction algorithms based on Rough Set have been summarized. The time complexity and space complexity of these algorithms have been analysed detailedly .The support oriented algorithm that proposed by Zhong is called Maximum Support Algorithm (MSA). The limitation of this algorithm is that it selects the preferable features that cause the highest support of the most significant rules rather than the highest overall quality of the potential rules. In other words, it only considers a local optimum instead of a global optimum of the potential rules. Moreover, sometimes this algorithm fails to make a choice between two sets of features when they have the same size of positive region and support of the most significant rule. This paper proposed a new algorithm called IMSA (Improved Maximum Support Algorithm. to resolve these shortcomings. We have defined a new heuristic function called Weight Support Heuristic. The experimental results show that this algorithm is effective in attribute reduction.A simple attribute reduction experiment system called RSS is developed. It realizes two main modules. The reduction module is to contrast the MSA and IMSA. The single module is to realize the concepts of Rough Set. Finally, the experimental results show that IMSA is more effective than MSA in attribute reduction of decision tables.