A multi-attribute group decision-making method that takes into account the disappointment of the subject was proposed for the multi-attribute group decision-making problem with distributed linguistic confidence evaluation information. In the actual group decision-making process the decision-makers have different levels of confidence in their own distributional linguistic evaluation information (which reflects the reliability of the distributional linguistic evaluation information) due to their background, experience, knowledge, etc. Based on this, a new evaluation structure was first defined, namely, the distributed language confidence evaluation. Secondly, under distributed linguistic confidence evaluation, the disappointment of a decision subject with bounded rationality is often implicit in the decision judgment or evaluation. Considering the above, the disappointment theory was used to incorporate the subject's disappointment into the decision-making process. Meanwhile, the subject's perceived utility under distributed linguistic confidence evaluation was obtained, and the subject's distributed linguistic confidence decision matrix was transformed into a subject's confident utility decision matrix. Then, the Dice similarity measure and the weighted Dice similarity measure for the confidence utility vector were defined. An optimization model was constructed to determine the attribute weights by combining the degree of incompatibility of the attributes. Next the alternatives were ranked on merit based on TOPSIS idea. Finally, the feasibility and effectiveness of the proposed method were illustrated by an example of decision-making in eco-industrial parks under the background of the circular economy.