文献类型：期刊文献

英文题名：A Golden-Section-Based group decision framework for software quality assessment under complex spherical fuzzy environments

作者：Yue, Chuan[1]

机构：[1]Guangdong Ocean Univ, Coll Math & Comp Sci, Haida Rd 1st, Zhanjiang 524088, Guangdong, Peoples R China

年份：2026

卷号：11

期号：7

起止页码：20815

外文期刊名：AIMS MATHEMATICS

收录：SCI-EXPANDED(收录号：WOS:001826426600001)、、Scopus(收录号：2-s2.0-105044725909)、WOS

基金：Acknowledgments This work was supported by the National Key R&D Program of China (No. 2024YFC3308004) .

语种：英文

外文关键词：group decision making; golden section; complex spherical fuzzy number; entropy; periodicity; software quality evaluation

外文摘要：This study introduces a novel group decision-making (GDM) framework under a complex spherical fuzzy environment, extending the Golden Section (GS) principle from a one-dimensional interval to a high-dimensional data center to address the core challenges of determining decision-maker (DM) weights and ranking alternatives. A rigorous theoretical justification is established, comprising four formal properties (self-similarity, optimal balance, convergence, and geometric center) that mathematically justify the GS ratio as aggregation weights for combining minimum and maximum matrices. An entropy-based method quantifies data quality for DM weight allocation, while a normalized Euclidean distance ranks the options. Eight experimental findings demonstrate the superiority of the GS-based approach: it enhances DM weight stability by a factor of 4.76 (376%) and improves ranking robustness by 8.33% compared to an arithmetic-mean-based center; against a direct extreme-value method, it achieves a 9.74-fold (874%) improvement in weight discrimination and a 0.42% sharper distinction among top alternatives. Dynamic experiments validate that the GS point 0.618 serves as a pivotal partition within [0, 1], with stable ordinal relations over 61.2% of the parameter space. Sensitivity analysis under +/- 50% weight variations (18 scenarios) confirms that the optimal alternative remains first in 100% of scenarios with minimal RC variability (std = 0.0013/0.0033). Comparisons with CSFN-TOPSIS and entropy-weighted methods show that the proposed method achieves the highest ranking discrimination (RDI = 0.8773). The proposed framework provides a robust, theoretically grounded methodology for reliable decision support in applications such as software quality assessment.