Considering the uncertainty of language assessment in decision makers, the interval value is used to indicate the language evaluation grade. Combining the Dempster-Shafer theory(DST) and data envelopment analysis(DEA), a hybrid multi-attribute decision making method based on the interval belief structure is proposed. Firstly, by using the interval cross efficiency model to construct the decision-making units interval belief structure of quantitative indicators, taking the evidence reasoning nonlinear optimization model to aggregate the interval belief structure, the interval belief structure of the quantitative indicators of decision-making units is converted into distributed evaluation vectors with interval belief according to the principle of utility(value) equivalence. Then, the interval belief structure of the quantitative and qualitative indicators is integrated through the evidence reasoning nonlinear optimization model, and the distributed evaluation vectors of the total interval belief structure of the decision-making units are obtained, and the expected utility theory is used to rank decision-making units. Finally, a numerocal example is given to illustrate the feasibility and effectiveness of the proposed method.