Dual-agent LLM extraction dataset: Taxonomy of sustainability innovation terms (2009–2026)
收藏资源简介:
This dataset contains the anonymized extraction outputs from a Systematic Literature Review (SLR) of 381 articles (2009–2026) examining eight sustainability innovation terms: eco-innovation, green innovation, sustainability-oriented innovation, sustainability-driven innovation, sustainability-related innovation, innovation for sustainability, innovability, and corporate social responsibility combined with innovation and sustainability strategy. Data extraction employed a dual independent large language model architecture (Claude Sonnet, Anthropic), with two agent instances processing each article independently against a 21-field structured instrument across seven thematic groups (identification, definitional content, inter-term relationships, theoretical foundations, methodological profile, contribution type, and conceptual synthesis). Inter-agent agreement for the primary classification field reached near-perfect agreement (Cohen's kappa = 0.950, n = 380). The dataset includes the consolidated extraction outputs, inter-agent concordance records, and corpus metadata. Analysis scripts are available from the corresponding author upon reasonable request. Associated article submitted to the Journal of Cleaner Production.



