Computational Discovery of Research Investment Patterns in Supply Chain Management: An Unsupervised Learning Framework for Funding Theme Extraction and Evolutionary Trajectory Analysis

Objective: This study maps global research investment patterns in supply chain management by identifying latent funding themes and their temporal trajectories. Methodology: This study analyzed ۷,۹۵۶ quality-assured grant records retrieved from Web of Science in June ۲۰۲۶, covering award years from ۱۹۶۳ to ۲۰۲۶. The framework combines systematic deduplication and quality filtering, multilingual transformer embeddings, K-means clustering with multi-metric validation, cluster-level TF-IDF, GERD normalization, right-censoring correction, and lifecycle sensitivity analysis. Results: The optimal solution contains ۱۳ thematic clusters. Five themes show GERD-normalized growth above ۶۷%: bioenergy value chains (۱۷۲%), quantum infrastructure (۹۵%), battery supply networks (۹۴%), additive manufacturing (۸۳%), and cybersecurity systems (۶۷%). Portfolio analysis assigns ۳۷.۶% of grants to emerging themes and ۷.۶% to mature themes, compared with the study benchmark of ۲۵% and ۳۰%, respectively. Conclusion: Grant-based science mapping provides forward-looking intelligence on supply chain research priorities. The results indicate rapid investment in sustainability and advanced technologies while suggesting a relative shortage of mature-stage translational research.

کلیدواژه ها:

Supply chain research investment ، grant portfolio analysis ، unsupervised machine learning ، research funding strategy ، temporal trajectory analysis