Article Preview
TopIntroduction
Decision-making trial and evaluation laboratory (DEMATEL) technique was first developed by the Geneva Research Centre of the Battelle Memorial Institute (Gabus & Fontela, 1972) to solve complex problems such as conflicting evidence (Zhang & Deng, 2019) and supplier selection (Liu et al., 2018). There are a number of approaches that can be adopted in order to make rational decisions, such as DEMATEL, AHP hierarchy process (Saaty, 1980), and technique for order preference by similarity to an ideal solution (known as TOPSIS) (Zavadskas et al., 2016). Among the above methods, DEMATEL has been widely used because of its large data capacity. DEMATEL is an effective method which analyzes the interrelationships between system factors and visualizes this structure through cause-effect relationship maps. It has been widely used in various areas (Altuntas & Gok, 2021; Büyüközkan & Güleryüz, 2016) and can be extended by other theories and approaches, such as grey decision making trial (Bai & Sarkis, 2013), analytic network process (Wu, 2008), and fuzzy numbers (Wu &d Lee, 2007). Until now, DEMATEL has been extended to make better decisions in different situations.
One of the important steps in DEMATEL is to normalize the direction influence matrix.Many typical DEMATEL methods exist. However, to the best of the authors’ knowledge, no normalized method from the perspective of maximum entropy (MaxEnt) in DEMATEL is available. From the informational perspective, it can be found that these existing normalized methods actually obtain different probabilities whose values are ranged from 0 to 1. Especially with regard to the row-sum normalized method, it obtains a probability distribution.
From a novel perspective of information theory, in this paper the authors propose DEMATEL based on MaxEnt, named as MaxEnt-DEMETEL. In information theory, entropy is used to measure the amount of information; the more uncertain the system is, the greater its entropy. MaxEnt is an effective tool for determining the amount of information existed in evidence. The proposed method applies the MaxEnt principle to the second normalized step of DEMATEL. Specifically, a novel normalized method is proposed in the MaxEnt-DEMATEL method. The process can be divided into three steps. Firstly, experts evaluate the direct relations of influential factors in emergency management. Evaluation results are presented in the form of intuitionistic fuzzy number (IFN). Secondly, the proposed method applies the MaxEnt principle to the second normalized step of DEMATEL. Thirdly, based on DEMATEL, the cause-effect classification of factors can be obtained. Finally, the cause factor is identified as the critical success factors (CSFs) in emergency management.
The proposed method has the following two advantages:
From the perspective of information theory, the MaxEnt principle is used to determine the normalized direct influence matrix (DIM), which allows to obtain the normalized matrix with minimal information loss, and thus ensures the reasonableness and accuracy of the DEMATEL results.
From the perspective of information theory, this normalization method has a clear physical meaning and is interpretable.
The rest of paper is organized as follows: In the next section, the authors introduce the preliminaries of this work; subsequently, they present the proposed method; then, they illustrate an application in emergence management; after this, they discuss the rationality and superiority of the proposed method; finally, they end the paper with the conclusion.