Abstract: Decision making trial and evaluation laboratory (DEMATEL) intended to study and resolve the complicated socioeconomic system issues, has been widely applied by domestic and foreign scholars in recent years. When analyzing the causal-relationship of complex socioeconomic system, a problem in the group decision making trial evaluation laboratory method is that experts may prefer to apply multi-granular evaluation scales to express their judgments. To conquer the problem, this paper applies 2-tuple linguistic model to express the experts’ preference judgment information, and proposes an improved method based on extended hierarchical linguistic model. The improved method constructs 2-tuple direct influence matrices with multi-granular evaluation information of group experts. It unifies the multi-granular information by extended hierarchical linguistic model. Then the unified information can be aggregated with the average operator of 2-tuple linguistic. Finally the traditional group DEMATEL method is processed for system information analysis and decision-making. An empirical study on internal barriers for automotive parts remanufacturers in China is presented to illustrate the feasibility of the proposed method.

Key words: decision making trial and evaluation laboratory, multi-granular evaluation scales, extended hierarchical linguistic model, 2-tuple linguistic model

摘要： 决策试行与评价实验室方法（decision making trial and evaluation laboratory，DEMATEL）作为一种面向复杂社会经济系统问题的因素分析算法，近年来受到国内外学者的广泛关注。为解决DEMATEL方法在分析复杂不确定社会经济系统时，由于专家知识背景不同而导致的群组专家选择多粒度评价标度进行偏好信息判断的问题，引入二元语义模型表达群组专家的偏好判断信息，提出一种基于拓展语言层级模型的多粒度群组DEMATEL改进方法。改进方法先对群组专家给出的不同粒度评价信息构造二元语义直接影响矩阵，再利用拓展语言层级模型进行归一化处理，并通过二元语义集合算子进行群组信息的集成，进而运用DEMATEL方法进行系统因素的因果关系分析与决策。最后，将改进方法应用于中国汽车零部件再制造的内部障碍因素分析，案例应用结果表明，所提方法在实践中是切实可行的。

关键词: 决策与试行实验室, 多粒度评价标度, 拓展语言层级, 二元语义模型