Abstract: Domain noumenon concept similarity had been widely used in medical diagnosis database，for overcoming the problems of low accuarcy in concept measurement and inadequte semantic fusion in domain concept similarity during the process of concepts interaction，proposes measurement theory and method of concept semantic noumenon similarity of medical domain. Be established in structure feature of domain noumenon，at first decomposes the semantic information contained in concept hiberarchy， and proposes computation model based on attribute， then determine the non-symmetric irreducible set in concepts in hiberarchy，through introducing the regulatory factor towards semantic model in sets，improves the fusion efficiency of domain concepts. Algorithm analy? sis and examples prove that the method effectively improve the accuracy of noumenon concept fusion.

Key words: Galois, breadth, asymmetric, semantici similarity