Scholarlink

<p>This study employs Bidirectional Encoder Representations from Transformers (BERT) to analyze the research landscape of the United Nations Sustainable Development Goals (SDGs). A dataset of 4951 articles from 1985 to February 2025 retrieved from Scopus was processed using BERT-based embeddings and clustering to identify thematic structures within sustainability-related scholarship. The analysis generated 33 initial clusters of semantically related keywords, which were refined into 20 thematic clusters and subsequently aggregated into five superclusters. These represent broad research domains: Marine &amp; Environmental Sustainability, Foundational Science &amp; Innovation, Regional Development &amp; Governance, Social Systems, Policy &amp; Food Security, and Global Energy &amp; Governance. The results demonstrate the interdisciplinary nature of SDG research, with strong attention to areas such as renewable energy, health governance, and food security, while also highlighting underexplored themes including demographic transitions, nano informatics, and biopolymer innovations. Methodologically, the study confirms the value of BERT in bibliometric mapping, offering more context-sensitive topic discovery than traditional approaches. Substantively, the outcomes provide an evidence-based framework for aligning research priorities with global sustainability challenges, supporting both scholars and policymakers in advancing the UN 2030 Agenda.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Can artificial intelligence decode the language of sustainability? Mapping global sustainable development goals research using BERT-based natural language processing

  • 期刊论文

  • 发表日期：

    2026年4月13日

查看全文
  • Chetan Sharma,
  • Shamneesh Sharma,
  • Meghna Luthra,
  • Rajender Kumar

摘要

This study employs Bidirectional Encoder Representations from Transformers (BERT) to analyze the research landscape of the United Nations Sustainable Development Goals (SDGs). A dataset of 4951 articles from 1985 to February 2025 retrieved from Scopus was processed using BERT-based embeddings and clustering to identify thematic structures within sustainability-related scholarship. The analysis generated 33 initial clusters of semantically related keywords, which were refined into 20 thematic clusters and subsequently aggregated into five superclusters. These represent broad research domains: Marine & Environmental Sustainability, Foundational Science & Innovation, Regional Development & Governance, Social Systems, Policy & Food Security, and Global Energy & Governance. The results demonstrate the interdisciplinary nature of SDG research, with strong attention to areas such as renewable energy, health governance, and food security, while also highlighting underexplored themes including demographic transitions, nano informatics, and biopolymer innovations. Methodologically, the study confirms the value of BERT in bibliometric mapping, offering more context-sensitive topic discovery than traditional approaches. Substantively, the outcomes provide an evidence-based framework for aligning research priorities with global sustainability challenges, supporting both scholars and policymakers in advancing the UN 2030 Agenda.