Abstract: An artificial immune recognition algorithm is proposed. Inspired by the metaphor of the interaction of antibodies with antigens, it processes the data as antigens, and then gets an antibody memory matrix by inspiring or restraining the activity of antibodies. The ability of data recognition and data structure description can be achieved by the memory matrix. The algorithm has the characteristics of recognition variety and self-adjustment. The experiment results demonstrate that the algorithm has good clustering effect and high recognition rate on R² space data and iris data.