| Formal concept analysis,first proposed by Wille in 1982,is a theoretical tool for data analysis and knowledge discovery.Formal concept analysis is original comes from the description of “concept” in philosophy,which is a unit consists of extent and intent.By concept derivation operator,the extent is expressed by a set of objects which possess common attributes,and the intent is a set of attributes which are commonly possessed by the objects in extent.Concept lattice,the core data structure in formal concept analysis,which shows the relationships between objects and attributes,can also describe the generalization and specialization relationships between different concepts.But for the concepts which are incomparable,there still exists some specific semantic can be explored.Considering this problem,the complementary concepts in concept lattice is proposed and defined formally,the relationship between incomparable concepts was studied under practical problems.The details are as follows:1.The complementary degree is introduced to describe the relation between two object subsets by considering a specific attribute subset,and according to different ranges of complementary degree,strong and weak complementary relations are defined respectively.2.The strong complementary concept is defined based on the strong complementary relation by considering a specific attribute subset,and the distribution average sum is introduced to explore the optimal strong complementary concept and corresponding generation algorithm;The semantic of strong complementary concept is weakened,and weak complementary concept and weak complementary concept based on a threshold are studied based on weak complementary relation.Furthermore,the decision rule and corresponding generation algorithm of the optimal weak complementary concept are given.Finally,examples are presented given to illustrate the value of complementary concepts proposed in practical problems such as finding cooperative teams and technical research teams. |
PDF Full Text Request