Attribute Reduction Based On Weighted Entropy And Two Subsets Of Concept Lattices

Rough set theory is a method of mathematics proposed by professor Pawlak in 1982 to deal with uncertain and imprecise knowledge.One of the main advantages of the theory is that it can handle information systems without any preliminary or additional information.Concept lattice theory,also known as formal concept analysis,is put forward by the German mathematician Wille in the same year.It is not only a kind of theory to conduct conceptual data and conceptual levels,but an effective tool for knowledge representation and knowledge discovery.One of the most core contents is attribute reduction problem in rough set and concept lattice theory.But due to the fact that the different application requirements and constraints,the existing reduction methods exist some limitations which lead to the reduction results maybe not meet users' requirements.So discussing more effective and more targeted reduction methods has greater practical significance.In this paper,the main works are as follows:1.In entropy reduction of rough set,users has prior knowledge of each attribute,such as the users' preference.However,this priori information does not reflect the classification ability of attribute set.Inversely,the importance of attribute only reflects the influence of the classification.So this article combines them by adding weight coefficient.Moreover,from the angle of information entropy,we redefine the attribute importance with weighted entropy,and propose weighted entropy reduction algorithm to gain attribute reduction which more meets the requirement of users.2.In the decision formal context,due to comprehension errors and measurement restrictions for data,some relationships between objects and attributes are uncertain.So dealing with incomplete decision formal context are more common.For incomplete decision formal context,this paper gives corresponding concept construction method which is consists of intension with double subsets,and comes up with a new method to structure concept lattice and corresponding improved algorithm.Furthermore,we put forward four judgement theorems of consistent sets and the method of attribute reduction.