Rescue factor for similarity measures of single-valued neutrosophic sets with insufficiencies

The similarity measure is an important indicator of the degree of similarity between two objects and is widely used in pattern recognition, medical diagnosis, and decision-making problems, etc. However, existing similarity measures based on single-valued neutrosophic sets (SVNS) have the weaknesses of insufficient applicability and interpretation, and fail to provide high credibility results. To address this problem, according to conventional ideas, new or improved versions of similarity measures should be constructed to demonstrate their superiority over existing methods. However, it has been increasingly challenging to implement this idea. In this paper, we provide a new idea by adding RFISM, a rescue factor, which can rescue most (some will be filtered out) of the similarity measures that have been shown to be insufficient in previous studies, while preserving the original methods. We implement the new idea by proposing SMS, a set of methods incorporating RFISM. We have compared and analyzed the methods in the SMS with the original methods in detail, and the results show that the selected original methods have all been rescued, with improved applicability and interpretation. Moreover, the new method provides high-confidence results in a specific instance of multi-period medical diagnosis.