The Attribute Reduction Of Information System Based On Concept Lattice

Rough set theory and formal concept analysis are useful tools for data analysis and knowledge discovery.Rough set theory use known knowledge to learn unknown or uncertain knowledge based on classification.The upper approximation and the lower approximation are key concepts of rough set theory.Operators in traditional rough set are constructed based on equivalent relationship and the application of equivalent relationship is limited,so the extension of rough set model has important theoretical significance and a high value of practical application.The traditional rough set is always discussed in single discourse.The information between object and attribute are not fully utilized.In this paper the single discourse is extended to double discourse.Two forms of new rough sets which are called object-oriented rough set and attribute-oriented rough set are produced.Their properties are discussed in detail.Based on the new models,the attribute reduction including shrinking reduction and extending reduction which can be used in information system and decision information system is studied.Concept lattice is the theoretical frame work in formal concept analysis in which the basic unit called concept run through the theory.It gives full expression of the relationship between object subset and attribute subset.Only one relationship between object and attribute is considered in the construction of tradition operators in concept lattice.In view of this shortcoming,new concept lattice including positive concept,negative concept and plus-minus concept are introduced by incorporating both discetnibility relation and indiscernibility relation into information system.Plus-minus concept lattice can describe objects and attributes from both positive and negative aspects efficiently.The attribute reduction of information system based on this concept lattice is investigated.Effectiveness can be proved from theory and simulated experiment.In order to describe the uncertainty and inaccuracy,the clear information is fuzzyfied.New concept lattice is constructed and the related properties are discussed in intuitionistic information system.In this information system,two pairs of operators based on positive relation and negative relation are introduced between attributes and objects.Three kinds of intuitionistic fuzzy concepts by introducing threshold value are proposed and the properties of intuitionistic fuzzy concepts are discussed.A new model of concept lattice is derived by defining dominance relation between objects and attributes and the issue on attribute reduction of intuitionistic fuzzy information systems is investigated under the principle of keeping Hasse graph invariant.Information system is an important mathematical model in the fields of artificial intelligence and pattern recognition.In this paper,rough communication between information systems is investigated and the model of lower approximation rough communication and upper approximation rough communication is produced and the related properties of the rough communication are investigated.Fuzzy rough set is generalization of rough sets in the framework of fuzzy set theory and fuzzy logic.In this paper,algebraic properties of generalized fuzzy rough approximations are investigated.Definable set is defined and the related properties are discussed.Necessary and sufficient conditions on information-preserving during communication are established.Two kinds of measures on characterizing lost information and distorted information are proposed to evaluate performance of the fuzzy rough communication.