TL;DR
This paper introduces the "Linguistic Picture Fuzzy Set" (LPFS) to better handle vague and imprecise decision-making information using linguistic terms. They then extend popular decision-making methods, TOPSIS and TODIM, to work with LPFSs, demonstrating their effectiveness with a numerical example.
What Problem Does This Paper Address?
Traditional decision-making often relies on precise numerical values, which can be limiting in real-world scenarios where information is inherently uncertain, vague, or requires nuanced expression. While fuzzy sets and their extensions like intuitionistic fuzzy sets (IFS) and picture fuzzy sets (PFS) have been developed to address this, they often struggle to fully capture complex human evaluations, especially when dealing with multiple response options (e.g., yes, abstain, no, refusal) and when decision-makers prefer using descriptive linguistic terms over numbers. This paper aims to bridge this gap by proposing a framework that combines the expressiveness of picture fuzzy sets with the naturalness of linguistic variables for multi-criteria decision-making.
Methods
The core of this paper lies in several key methodological contributions:
- Linguistic Picture Fuzzy Set (LPFS): This is the central concept. LPFS extends the idea of Picture Fuzzy Sets by representing the positive, neutral, and negative membership degrees using linguistic variables (terms from a linguistic term set). This allows for more intuitive and human-like representation of preferences.
- LPFS Operations: The authors define operational rules for LPFSs, including aggregation operators like the Linguistic Picture Fuzzy Weighted Averaging (LPFWA) and Linguistic Picture Fuzzy Weighted Geometric (LPFWG) operators.
- Distance Measures: To quantify the differences between LPFSs, they introduce generalized weighted distance, generalized weighted Hausdorff distance, and generalized hybrid weighted distance measures. These measures are designed to be robust and capture various aspects of LPFS differences.
- Entropy Weighting: To objectively determine the importance of different criteria, they propose a linguistic picture fuzzy entropy method. This is then used to calculate criterion weights.
- TOPSIS and TODIM Extensions: The authors adapt the well-known TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution) and TODIM (interactive and multi-criteria decision-making) methods to work with LPFS information, incorporating the proposed distance measures and entropy-based weights.
Datasets
The paper does not use any external datasets for training or evaluation. Instead, it introduces a numerical example (adapted from Chan and Kumar [42]) to illustrate the application and effectiveness of the proposed methods. This example involves a supplier selection problem with four suppliers evaluated across five criteria, represented using linguistic terms.
Key Insights and Ideas
- Bridging the Gap: LPFS effectively combines the expressiveness of picture fuzzy sets with the intuitive nature of linguistic variables, offering a more natural way to represent complex decision-making information.
- Robust Decision-Making: Extending TOPSIS and TODIM to LPFSs, coupled with entropy-based weighting and novel distance measures, provides a more comprehensive and potentially more accurate approach to multi-criteria decision-making under uncertainty.
- Sensitivity Analysis: The study demonstrates that the proposed methods are relatively stable with respect to parameter variations in the distance measures and decision-maker preferences, suggesting their practical applicability.
Key Results
The paper demonstrates the feasibility and effectiveness of its proposed methods through a numerical example:
- TOPSIS Method: The TOPSIS method, using the proposed LPFS framework and entropy weights, identified supplier A3 as the best alternative, with a ranking order of A3 > A4 > A1 > A2.
- TODIM Method: The TODIM method, applied to the same problem, also ranked A3 as the best alternative, with the same ranking order: A3 > A4 > A1 > A2.
- Comparative Analysis: The results from the proposed TOPSIS and TODIM methods were compared with existing approaches (e.g., using different distance measures or aggregation operators). While most methods yielded similar rankings, the paper highlights potential issues with other aggregation operators (like LPFWG) that might lead to different results or loss of information. The proposed methods showed robustness across different parameter settings.
Limitations and Future Work
The authors acknowledge several limitations:
- Integer Subscripts: The current linguistic term sets are based on integer subscripts, which might lead to a loss of evaluation information.
- Aggregated Operators: The aggregated operators might not always result in a linguistic term set, which could be counterintuitive.
- Hausdorff Distance: The hybrid distance measure's reliance on the maximum value (Hausdorff distance) might make it susceptible to extreme values, potentially leading to inaccurate rankings in certain scenarios.
Future work suggested includes:
- Linguistic Scale Functions: Incorporating linguistic scale functions for different semantic situations and extending to uncertain linguistic term sets.
- New Aggregated Operators: Developing new aggregated operators that guarantee closure within the linguistic picture fuzzy set framework.
- Enhanced Distance Measures: Extending hybrid distance measures to consider the mean and variance of LPFSs and exploring their application to continuous distance measures.
- Practical Applications: Applying the proposed methods to real-world problems like pattern recognition and medical diagnosis.
Summary updated