To deal with the problem of supplier selection decision-making with attribute evaluation value as linguistic variables and unknown expert weights, this paper proposes an expert weight determination method that comprehensively considers the hesitancy degree and similarity of evaluation information. Firstly, the linguistic variable conversion standard is improved based on the difference in the hesitation degree of expert evaluation, and the linguistic variables are converted into intuitionistic fuzzy numbers that is more in line with the actual decision-making. Then, the expert weights are integrated from the hesitation degree and similarity of the evaluation information to obtain the aggregated synthesis evaluation matrix. Finally, the TOPSIS method is used to sort suppliers. The example verifies the feasibility and effectiveness of the proposed method. The results of sensitivity analysis and comparative analysis show that the different preference of a decision maker on the certainty and consistency of expert evaluation will affect the final decision-making result. When experts have different hesitations in attribute evaluation due to cognitive limitations and personal preferences, the linguistic variable conversion method taking into account the differences in hesitation can reduce the unreasonable impact of evaluation information uncertainty on evaluation results, which is helpful to improve the credibility of the evaluation result.