Abstract
This research proposes a novel score function for Intuitionistic Fuzzy Sets (IFSs) based on the Poincaré metric from hyperbolic geometry. While Fuzzy Sets (FSs) offer flexibility in decision-making, they often result in nondecisive outcomes when membership degrees are similar. By incorporating the hesitation degree alongside membership and non-membership degrees, IFSs provide a more comprehensive data structure. The proposed approach expresses these degrees as two-dimensional vectors within a Poincaré metric framework. This score function exhibits strict monotonicity, ensuring clear preferences without the need for secondary accuracy functions. Comparative analysis demonstrates that the proposed function successfully discriminates between candidates where existing methods fail, providing a solid foundation for hesitation analysis in complex decision-making problems.
Introduction
In the field of decision science, uncertainty is a constant challenge. Researchers often employ Fuzzy Sets to model vague information. Intuitionistic Fuzzy Sets (IFSs) represent a significant advancement by capturing three critical dimensions: membership (support), non-membership (opposition), and hesitation (uncertainty).
However, a major bottleneck exists in the "score functions" used to rank these sets. Many current functions lack sufficient discrimination, often assigning the same score to different fuzzy sets. This necessitates the use of a secondary "accuracy function" to break ties, which can be mathematically cumbersome. Furthermore, existing models frequently struggle to integrate the "hesitation" degree effectively, leading to indecisive rankings when membership and non-membership values are closely aligned.
Methodology: The Poincaré Metric Approach
To overcome these limitations, this paper introduces a score function ($S_P$) leveraging the Poincaré metric, a cornerstone of hyperbolic geometry. This geometric approach moves beyond simple arithmetic subtraction to a more robust distance-based evaluation.
Vector Representation and Hyperbolic Mapping
Instead of treating membership ($\mu$) and non-membership ($\nu$) as simple scalars, the authors represent them as two-dimensional vectors. These vectors are mapped into a hyperbolic space—specifically the Poincaré half-plane. The score is then derived using the arc hyperbolic cosine ($acosh$), which naturally accounts for the internal structure of the fuzzy data.
The Role of Hesitation
A key innovation of this method is how it treats the hesitation degree ($\pi$). In the geometric model, hesitation acts as a significance modifier. As illustrated in the following figure, the inclusion of hesitation transforms point coordinates into segments, effectively amplifying the differences between alternatives based on the level of uncertainty.
Strict Monotonicity
The paper mathematically proves that the $S_P$ function is strictly monotonic. This means that any marginal increase in membership or decrease in non-membership will always result in a distinct, predictable change in the final score. This property is crucial as it eliminates the need for tie-breaking functions and ensures that all attributes are properly reflected in the final ranking.
Key Insights and Results
Sensitivity and Discrimination
Because the Poincaré metric has an exponentially increasing rate, it is far more sensitive to slight changes in data than traditional linear functions. This makes it an effective "magnifying glass" for subtle differences between candidates. In comparative tests against established methods (e.g., Chen & Tan, Gao et al.), the proposed $S_P$ function successfully discriminated between all candidates, even in cases where existing methods produced identical scores or encountered mathematical errors like zero denominators.
Real-World Application: LSRPV Site Selection
The model's efficacy was further validated through a case study involving the site selection of large-scale rooftop photovoltaic (LSRPV) projects across 10 Chinese cities. Evaluating 15 criteria across economic, social, and meteorological categories, the method identified Hefei ($A_3$) as the top choice. The results showed a strong positive correlation ($\rho = 0.803$) with existing state-of-the-art methods, confirming the model's reliability while offering more granular and distinct scoring gaps.
Conclusion
The application of the Poincaré metric to Intuitionistic Fuzzy Sets provides a superior framework for decision-making under uncertainty. By explicitly addressing hesitation and ensuring strict monotonicity, the proposed score function offers a more discriminative and reliable ranking mechanism. Future work will focus on developing objective fuzzification methods to reduce human bias when converting raw data into fuzzy sets, further enhancing the applicability of this framework to automated and AI-driven decision systems.
Summary updated