A method based on the attribute set of target case is proposed to retrieve historical cases from the case warehouse, so as to get the qualified historical case to form a sub-case warehouse. Then, a G1 and Kernel Density based group decision method is used to allocate the weights of the target case’s attributes. Considering the features of numerical attributes and symbolic attributes, the corresponding similarity functions are used to calculate the similarities of cases under different attributes. Thus the similarities between historical cases ...