The Research Of Rough Sets Reduct Based On Concept Lattice

Reduct and core are the hotspots in rough sets theory. The existed reduct algorithms have following problems: they can't make sure of the completeness of result and can't assure to obtain the smallest reduct;they often need large space, and have bad performance.Concept lattice describes the relation of generalization-specialization between concepts and the relationship between intension and extension. It is a complete form of knowledge representation. Both concept lattice and rough sets are based on equivalence calss, so the compact relation exsits between them.This dissertation discusses and researches the relationship between concept lattice and rough sets, then presents the resolution of reduct and core based on the concept lattice. The content of the dissertation is as follows:Related works on concept lattice and rough sets theory are stated generally.The representation of rough sets based on concept lattice is discussed in detail, and relevant properties are presented.The reduct algorithm of information system and decision table based on concept lattice is presented as ARBL. This algorithm can process attribute reduct and value reduct, and it can get all the reducts.Simplified concept lattice is introduced to the process of reduct resolution, and the construction algorithm SimpG is presented. Simplified concept lattice reduce the scale of original concept lattice through cutting the nodes irrelevant to the reduct and core, as a result, the efficiency of constructing concept lattice is greatly improved.Based on simplified concept lattice, the algorithm of reduct resolution based on simplified concept lattice is presented as RBSL, this algorithm can get the core, all the reducts, and the consistency of decision table at the same time.Based on the researches stated above, the prototype system of reduct resolution based on concept lattice is implemented.