Attribute Reduction Of Rough Set Based On Conditional Information Entropy In General Binary Relation

The rapid development of information technology to accelerate the advent ofglobalization of information, using information tools to enhance its competitive advantage hasbecome the means of many enterprises to adopt, a large amount of data accumulated in thecourse of business, how to deal with the analysis of these data and gain valuabletheinformation is an important research topic in data mining. Rough set theory is an importanttool in data mining, mainly deal with uncertainty and incomplete knowledge of the data. Thetheory proposed by Pawlak, the core concepts of the classical rough set theory is based on theequivalence relation on approximation and lower approximation, these two concepts on thebasis of the potential knowledge of information systems can be expressed as a form ofdecision rules. Equivalence relation constraints are too harsh, limiting the application of theclassical rough set theory in complex information systems, so many researchers extend theequivalence relation for the compatibility relations, similarity relations and binary relations.Information entropy can measure the degree of uncertainty of the event, to be able to usethe exact value to measure the uncertainty of knowledge. Therefore, through theestablishment of the relationship between rough set theory and information entropy, theinevitable roughness of the knowledge to make a more accurate measure, and thus moreaccurate rough set attribute reduction, useful rules.Many researchers in the information entropy theory is applied to measure properties ofthe classical rough set knowledge roughness reduction done a lot of work. In recent years,some researchers equivalence relation expanded into a general binary relation, and theinformation entropy theory application of them. But the application of information entropybased on general binary relations only stay in the rough knowledge level on the basis of theresults of previous studies, the application of information entropy theory to the general binaryrelation in rough sets attribute Reduction above, this theory is applied to broaden the researchroad.On the basis of the above study, the definition and nature of knowledge based on generalbinary relations on conditional information entropy, and prove some theorems related basedon the description based on the concept of equivalence relations on conditional informationentropy. Use existing reduction algorithm based on conditional information entropy based ongeneral binary relations, proposed an improved rough set attribute reduction algorithm, and apply it to the processing of plant data, confirmed the improved after the advanced nature ofthe algorithm.