Using mutually validated memories of experts for case-based knowledge systems

摘要

Organizing a reliable case base, which serves as a repository of experience, is crucial for the success of a case-based reasoning (CBR) system. To ensure that such repositories contain high-quality cases, this paper proposes a framework employing the methodology of fuzzy linguistic group decision-making (GDM) in the context of multiple attributes. The overall process of MAGDM could be analogous to the memory-related behaviors of the human brain, in which knowledge is elicited and validated, as in the short-term memory, and then eventually integrated into the long-term memory to serve as solutions to build-up the number of high-quality cases. Moreover, the proposed approach is flexible, as it enables experts to define the set of the parameters of the membership functions associated with labels, thus improving the quality of the linguistic term sets and leading to better assessments. Furthermore, the proposed KC index, characterized by measures of both individual and group consistencies, can provide a more effective assessment to assign suitable experts’ weights than most existing GDM models. This is supported by the experimental results presented in this work, indicating that the KC index can indeed lead to a more satisfactory overall level of consensus. In addition, the mutual validation between the set of the parameters of the membership functions associated with labels by experts and the evaluation of the experts’ weights can be manifested in terms of the KC index.The extended collective decision matrix derived from the process of MAGDM that is used to construct case bases is more practical and effective than other approaches, as its elements are meaningful and interpretable. The proposed framework, integrating fuzzy linguistic GDM and CBR, thus enhances the efficiency and effectiveness of a CBR system. This is further evidenced in the results of an experiment, which show that this hybrid framework is very effective in implementing a case-based knowledge system and provides a powerful methodology for performance ranking.