The Construction Of IS-concept Lattice In Information Table

Concept lattice,also known as Galois lattice,is a concept hierarchy structure based on the binary relationship between objects and attributes in datasets.Each concept consists of two parts: intension(that is,the description of concepts)and extension(that is,the objects covered by connotation).The process of generating concept lattice from formal context is essentially a process of concept clustering.In addition,concept lattice vividly and concisely reflects the generalization and specialization of these concepts through Hasse diagram.It has been widely used in knowledge engineering,machine learning,pattern recognition,expert system,computer network,data analysis,decision analysis,data mining and other fields.In the research of concept lattice,each attribute in the context usually takes only two values,so for the information table with multiple attribute values,we usually use the positive scaling method to transform it into a formal context,and then construct the concept lattice from the derived formal context.This thesis attempts to constract concept lattice directly from an information table.The main resules are as follows:(1)For a complete information table,by defining the maximum description formula of an object set and the meaning set of a formula,the IS-concept lattice is constructed.And the algorithm for finding all the IS-concepts from the complete information table and the attribute reduction of the IS-concept lattice are formulated.(2)For incomplete information table,since the values of some objects under some attributes are uncertain,there are some differences in the definition of concept.The meaning set of a formula is defined as an interval set of object.By defining the maximum description formula of object interval set and the meaning set of formula,the IS-concept lattice from incomplete information table is constructed,and the algorithm for finding the attribute reduction of IS-concept lattice from incomplete information table.(3)The relationship between concept lattice in formal context and IS-concept lattice in information table is given.