A hybrid DEMATEL-FRACTAL method of handling dependent evidences. (May 2020)

• Record Type: Journal Article
• Authors: Mao, Shengzhong; Han, Yuzhen; Deng, Yong; Pelusi, Danilo
• Is Part Of: Engineering applications of artificial intelligence. Volume 91(2020)
• Publication Date: 2020-05
• Subjects: Dempster–Shafer evidence theory -- Dependence -- DEMATEL -- Discounting evidence -- Fractal -- Decision-making; Engineering -- Data processing -- Periodicals; Artificial intelligence -- Periodicals; Expert systems (Computer science) -- Periodicals; Ingénierie -- Informatique -- Périodiques; Intelligence artificielle -- Périodiques; Systèmes experts (Informatique) -- Périodiques; Artificial intelligence; Engineering -- Data processing; Expert systems (Computer science); Periodicals; 620.00285
• DOI: 10.1016/j.engappai.2020.103543
• Languages: English
• ISSNs: 0952-1976

Abstract

Abstract: Dempster combination rule in evidence theory is widely used in data fusion system. One assumption of Dempster combination rule is the independence among different evidences. However, it is difficult to satisfy this requirement due to many influencing factors. The main contribution of this paper is to propose a systematic model to deal with dependence evidences in evidence theory. The core of the model can be divided into two parts: handling inner dependence and handling outer dependence. For the inner dependence, we use DEMATEL model to establish the relationships between different components in the system and get their relative weights considering their influences. For the outer dependence, we use BPAs inherent fractals features to deal with the uncertainty in the outer environment as well as the dependence among each collected evidence. After that we combine these two aspects of dependence and make decisions. By discounting evidences we avoid redundant calculation thus obtain useful information as much as possible. A case study of transportation project selection problem is used to illustrate our proposed method. Highlights: A DEMATEL method to describe the inner dependence of the components in a system. Implement essential fractal features of belief functions to solve outer dependence. The ability to tackle with uncertain information is strong. Discounting belief functions to avoid the effects exerted by dependence among them.