Abstract: The data mining of incompatible information systems is difficult, and it is one of the most important research directions of classical data mining.This paper studies the structure of granular computing and the model of computation based on the classical rough set and proposes a method for the data mining of incompatible of information systems based on granular matrixes.Firstly, the granular matrix which saves redundant data,the coordination degree algorithm and the controlled conductivity of granular matrixes are defined. Secondly, the attribute reduction algorithm of incompatible information systems is designed on the basis of the above theory.The final simulation experiments and preliminary enterprise application show the robustness and availability of the algorithm .This study provides an effective granular computation model for the data mining of incompatible information systems.

Key words: rough set, information granule, granular computing, matrix