A consistency analysis-based rule activation method for extended belief-rule-based systems

作者：Yang, Long-Hao; Wang, Ying-Ming^*; Fu, Yang-Geng

来源：Information Sciences, 2018, 445: 50-65.

DOI：10.1016/j.ins.2018.02.059

• Activation framework
• Consistency analysis
• Extended belief-rule-base
• Rule activation
• Selection model

Problems with inconsistency and incompleteness are widely found in rule-based decision support systems. These problems often impact the accuracy and usability of rule-based decision support systems. The present work focuses on an advanced rule-based decision support system, namely the extended belief-rule-based (EBRB) system, and proposes the consistency analysis-based rule activation (CABRA) method to overcome the above two problems simultaneously. However, two challenges must be discussed and addressed for the EBRB system. First, rather than using activated weights, suitable activated rules must be redefined to better analyze the problems of inconsistency and incompleteness. Second, suitable activated rules must be selected without having to depend on subjective information. Therefore, the proposed CABRA method uses the set of consistent rules as an activation framework to define suitable activated rules before calculating their activated weights, and utilizes the CCR model as a selection model to select suitable activated rules from the set of consistent rules. As such, by embedding the CABRA method, the EBRB system can overcome the problems of inconsistency and incompleteness. Three case studies demonstrate how the use of the CABRA method improves the accuracy and rule activation rate of the EBRB system, which is further confirmed by comparisons with the results of other existing studies. In addition to the work performed in the EBRB system, the CABRA method is treated as a generic rule activation method that can be available for other rule-based decision support systems.

• 出版日期2018-6
• 单位福州大学