Inter observer variability assessment through ordered weighted belief divergence measure in MAGDM

Numerous multi-attribute group decision-making (MAGDM) methods have been proposed to achieve consensus results. However, most existing approaches neglect the inter-observer variability among experts' opinions, focusing solely on equal or varying priority assignments. Ignorance of evidence supporting the expert's judgment for alternative leads to an uncertain decision and prevents a reliable decision. Therefore, this study aims to propose an Evidential MAGDM method by assessing inter-observational variability and handling conflicts that emerge among experts. The proposed method is presented in three steps. First, a basic probability assignment generation method is introduced to quantify the inherent characteristics of each alternative and subjective opinion. Second, the ordered weighted belief and plausibility measures are introduced to quantify the true nature of alternatives. It assesses how strong evidence is supporting experts' opinions for each alternative, simultaneously addressing inter-observational variability and conflicts that emerge among the group of experts. Third, an ordered weighted belief divergence measure is constructed to obtain consensus evidential weights for each expert, thereby establishing an order of preferences among the experts. The proposed method's reliability is validated through two case studies: a human resources selection problem and a deep neural network based classification task for diagnosing retinal disorders using optical coherence tomography images.