Elsevier

Information Sciences

Volume 102, Issues 1–4, November 1997, Pages 111-131
Information Sciences

Intelligent system
An extended rule-based inference for general decision-making problems

https://doi.org/10.1016/S0020-0255(96)00215-0Get rights and content

Abstract

Complex decision-making problems are usually modeled in terms of a number of decisive variables that are related hierarchically. Pieces of evidence are aggregated in a bottom-up way to determine the final decision goal. For dealing with such problems, Ishizuka et al. [5,6] developed a rule-based inference procedure by using fuzzy sets to represent imprecise data, and employed the Dempster-Shafer theory to aggregate evidence for structural damage assessment. This paper highlights the necessity of applying the rules with all possible consequent clauses in the procedure with the considerations of information quality and quantity, when applying the procedure to more general decision-making problems. The use of such rules can provide high-quality information, which contains much less ignorance in the resultant information. Also, it can afford more information due to the decrease in total ignorance. A simple example with two cases is used to compare the results based on the two different presentations of rules in the inference procedure for illustrating the importance of considering the rules proposed in this paper.

Cited by (17)

  • Belief rule-based classification system: Extension of FRBCS in belief functions framework

    2015, Information Sciences
    Citation Excerpt :

    The belief functions theory, also called Dempster-Shafer theory, proposed and developed by Dempster [13] and Shafer [46] et al., has become one of the most powerful frameworks for uncertain modeling and reasoning. As fuzzy sets theory is well suited to dealing with fuzziness, and belief functions theory provides an ideal framework for handling imprecision and incompleteness, many researchers have investigated the relationship between fuzzy sets theory and belief functions theory and suggested different methods of integrating them [6,7,34,56,57]. Among these methods, Yang et al. [57] extended the fuzzy rule in belief functions theory and proposed a new knowledge representation scheme in a belief rule structure, which is capable of capturing fuzzy, imprecise, and incomplete causal relationships.

  • Maximal confidence intervals of the interval-valued belief structure and applications

    2011, Information Sciences
    Citation Excerpt :

    The DS theory is of fundamental importance in artificial intelligence and cognitive science. It has been applied in a wide range of areas including information fusion [7,74,82,96], expert systems [1,13,18,75,92], pattern recognition and classification [15–17,23–25,30,48,70], diagnosis and reasoning [10,21,36–38,51,52,54,55,69], knowledge reduction [83–85], decision analysis [8,11–14,46,74,80,86,88,90,93–95], intelligent control [33], process monitoring [71–73], audit risk assessment [31,32,40,41,63–68], environmental impact assessment [79], image processing [9,19,43,44,49,56], geographic information system [47], contractor selection [61,62], organizational self-assessment [58,91], safety analysis [45,78] and regression analysis [50,53]. In general, applying DS involves three fundamental steps: first, formulate mass functions (or basic belief assignments, BBAs) to represent uncertainties existing in the problem; second, compute belief functions on the basis of the mass functions; and finally, if necessary, combine mass functions into a single function.

  • On the dynamic evidential reasoning algorithm for fault prediction

    2011, Expert Systems with Applications
    Citation Excerpt :

    Dempster–Shafer (D–S) theory of evidence. Due to the power of the D–S theory in handling uncertainties (Bauer, 1997; Beynon, Cosker, & Marshall, 2001, 2002a; Chen, 1997; Yager, 1992; Yen, 1990), so far, it has found wide applications in many areas such as expert systems (Biswas, Oliff, & Sen, 1988; Beynon et al., 2001; Chen, 1997; Wallery, 1996), uncertainty reasoning (Benferhat, SafHotti, & Smets, 2000; George & Pal, 1996; Hullermeier, 2001; Ishizuka, Fu, & Yao, 1982; Jones, Lowe, & Harrison, 2002; Rakar, Juricic, & Ball, 1999), pattern classification (Binaghi & Madella, 1999; Binaghi, Gallo, & Madella, 2000; Denoeux, 1997, 1999, 2000a, 2000b; Denoeux & Zouhal, 2001; Denoeux & Masson, 2004), fault diagnosis and detection (Fan and Zuo, 2006a, 2006b; Parikh and Pont, 2001; Rakar and Juricic, 2002; Yang and Kim, 2006), information fusion (Fabre et al., 2001; Ruthven and Lalmas, 2002; Telmoudi and Chakhar, 2004), Multiple attribute decision analysis (Beynon et al., 2001; Beynon, 2002a, 2002b; Guo, Yang, & Chin, in press; Xu, Yang, & Wang, 2006; Yang & Xu, 2002a, 2002b; Yang & Sen, 1994; Yang, Liu, Wang, Sii, & Wang, 2006a; Yang, Wang, Xu, & Chin, 2006b), image processing (Bloch, 1996; Huber, 2001; Krishnapuram, 1991) and regression analysis (Monney, 2003; Pent-Renaud and Denoeux, 2004; Wang and Elhag, 2007). In the last decade, an evidential reasoning (ER) approach has been developed for MADA under uncertainty (Xu & Yang, 2003; Xu et al., 2006; Yang & Sen, 1994; Yang & Xu, 2002a, 2002b, 2004; Yang et al., 2006a, 2006b).

View all citing articles on Scopus
View full text