Muhammad Jauharul Fuady, Aji Prasetya Wibawa, Dika Aurelya Aleandra Taroreh, Adelia Miftakul Janah, Anissa Alifia Putri
Sustainable Development Goal mappings provide structured, query-based frameworks for linking scholarly publications to global sustainability targets. The Elsevier SDG Mapping dominates this domain, yet its evolving Boolean queries introduce hidden label drift that necessitates frequent reprocessing of publication corpora as a computationally prohibitive task for large-scale bibliometric datasets. This study presents a lightweight diff-based methodology to detect partial updates across five releases (2019-2025), enabling selective reprocessing only when definitional changes occur. Pairwise comparison of 80 query pairs reveals > 50% of version transitions modify fewer than half the SDGs (e.g., 2023-2025: 7 / 16 changed). Syntax normalization and systematic differencing capture subtle changes, ensuring reclassification targets exclusively affected publications. The resulting workflow reduces redundant computation by > 8 0% for minor updates without compromising accuracy. Identification of persistent datapoints surviving definitional updates offers future pathways for high-confidence corpora. Version-aware query differencing maintains reproducibility, efficiency, and transparency in sustainability research analytics. © 2025 IEEE.
Universitas Negeri Malang, Electrical and Informatics Engineering, Malang, Indonesia
