As a service oriented architecture

data grid provides distributed data access

storage and management for remote users.It has become growing concern among developers about the data classification issue in data grid environment.This paper proposes a high-efficient system used in data grid which makes dynamical synthesis of multiple methods(DSMM)

distributes workloads to each node of the grid with the load-balance premise

and improves the low classification accuracy points of traditional methods.In spite of that

as an application in loose coupling grid environment

DSMM

which provides life time management of multiple classification methods

guarantees its robustness and flexibility.Experiment uses 2927 breast cancer cases and proves that the DSMM system has flexibility

high efficiency and increases the classification accuracy in data grid.