Mapping Convergence Themes in Web of Science A Thematic Evolution Analysis
1
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
This study maps convergence in engineering using a reproducible single-parameter pipeline applied to 948 Web of Science Core Collection records published between 2019 and 2024. The analysis relies on the KW_Merged field (Author Keywords ∪ Keywords Plus) and jointly estimates a keyword co-occurrence network, a centrality–density strategic map, and a two-period thematic evolution (2019–2021 → 2022–2024) under association-strength normalization, Louvain community detection, and a Fruchterman–Reingold layout. The intellectual structure consolidates into three blocks—method, system/design, and applications—with convergence brokering across them. Strategic-map evidence shows persistent core hubs, while peripheral micro-themes emerge and reposition along an AI- and sustainability-oriented axis, indicating a gradual shift from computation-centric topics (modeling, simulation, optimization) toward data-intensive intelligence and sustainable systems (machine learning, computer vision, deep learning, energy efficiency, smart grids). Thematic evolution highlights the direct continuity of the algorithm–convergence–design axis (inclusion ≥ 0.30), the inflow of consensus protocol and multi-objective optimization as routes of application-level expansion, and a relative decline of finite-element and voltage-control families toward the margins. Interpretations remain stable under compact threshold checks (minimum occurrence 3/5/7; adjusted inclusion cutoffs), supporting robustness.
Abstract
This study maps convergence in engineering using a reproducible single-parameter pipeline applied to 948 Web of Science Core Collection records published between 2019 and 2024. The analysis relies on the KW_Merged field (Author Keywords ∪ Keywords Plus) and jointly estimates a keyword co-occurrence network, a centrality–density strategic map, and a two-period thematic evolution (2019–2021 → 2022–2024) under association-strength normalization, Louvain community detection, and a Fruchterman–Reingold layout. The intellectual structure consolidates into three blocks—method, system/design, and applications—with convergence brokering across them. Strategic-map evidence shows persistent core hubs, while peripheral micro-themes emerge and reposition along an AI- and sustainability-oriented axis, indicating a gradual shift from computation-centric topics (modeling, simulation, optimization) toward data-intensive intelligence and sustainable systems (machine learning, computer vision, deep learning, energy efficiency, smart grids). Thematic evolution highlights the direct continuity of the algorithm–convergence–design axis (inclusion ≥ 0.30), the inflow of consensus protocol and multi-objective optimization as routes of application-level expansion, and a relative decline of finite-element and voltage-control families toward the margins. Interpretations remain stable under compact threshold checks (minimum occurrence 3/5/7; adjusted inclusion cutoffs), supporting robustness.