A Novel Weighted Evidence Combination Method Based on Reliability and Credibility degree with an Application in Decision Making Process

The Dempster–Shafer theory (DST), also known as Belief function theory (BFT), finds extensive application across various domains involving data fusion. However, it is not uncommon to encounter counterintuitive results when combining highly conflicting pieces of evidence using the Dempster combination rule (DCR). While the literature offers several combination rules, only a few address these shortcomings effectively. To address these limitations and rectify the issue of counterintuitive outcomes, a novel weighted evidence fusion rule has been introduced. This fusion approach comprises four key steps: First, the reliability degree of each evidence set, also known as body of evidence (BOE) is determined based on its compatibility degree and total correlation. This step helps to mitigate conflicts arising from unreliable sources. Second, the degree of credibility of each BOE is computed, taking into account distance of evidence and uncertainty degree, further enhancing the accuracy of the fusion process. Third, the reliability and credibility degrees are combined to derive weights for each BOE, ensuring that more weight is assigned to trustworthy sources of evidence. Finally, these weights for each BOE are used to compute a modified Basic Probability Assignment (BPA) for each proposition, which are then fused using the Dempster combination rule. Numerous numerical examples, as well as artificial and real datasets, have been employed to emphasize the effectiveness and potential of this innovative approach. Furthermore, through comparative analysis, it becomes evident that the proposed rule consistently outruns other existing methods, ensuring more accurate and reliable outcomes.