Aiming at the defects of TOPSIS (technique for order preference by similarity to ideal solution), two improved TOPSIS methods based on Tanimoto coefficient and symmetric difference are proposed. Improve or solve TOPSIS index correlation problem, special sample set can not compare the advantages and disadvantages of the problem and the sample data dynamic changes in the reverse phenomenon and other defects; The classical TOPSIS model, the improved Tanimoto model and the improved symmetric difference model were compared and verified in terms of stability, specificity, sensitivity and validity, and the application scenarios of the two improved models were given. The results show that the two methods have their own advantages.