Abstract: Aiming at the complex multi-attribute decision-making problem with natural language scales and subjective scores, by combining the cloud model theory, a TOPSIS method based on natural language cloud is constructed. Firstly, nonlinear programming is designed to improve the calculation in entropy, and hyper-entropy is obtained through the cloud atomization property, so that the natural language scales are transformed into cloud models. Secondly, the subjective scores are transformed into evaluation clouds to obtain the decision-making cloud matrix, and thus the multi-attribute evaluation method is established by the definition of TOPSIS. Finally, the proposed method is applied to the evaluation of logistics park location schemes. The effectiveness of the proposed method is verified by compar- ing it with the method in existing literatures and traditional TOPSIS method. 

Key words: cloud model, multiple attribute decision making, natural language, TOPSIS