Sustainable Procurement Disclosure Index (SPDI) Assessment Dataset: Human and AI-Generated Scores for Indonesian Listed Firms
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
This dataset supports a study on human–AI collaboration for sustainability report assessment using the Sustainable Procurement Disclosure Index (SPDI). It contains four files: (1) the SPDI Guideline, which defines 33 GRI-based indicators across governance, stakeholder engagement, environmental, social, and economic categories, each with a 0–4 scoring rubric; (2) human expert (ground truth) scores for 99 firms listed on the Indonesia Stock Exchange (IDX), assessed across all 33 SPDI indicators; (3) AI-generated scores produced by a Gemini LLM-based system using structured prompt engineering against the same 33 indicators for the same 99 firms; and (4) a comparison file containing per-company and per-indicator similarity indices between AI and human assessments, along with industry-level aggregations. The sustainability reports evaluated were publicly available 2023 annual/sustainability reports. The data enables research on automated disclosure assessment, inter-rater agreement between human experts and LLMs, and the broader applicability of AI in corporate sustainability evaluation.
Fields of Study & Subjects
Corporate Social Responsibility Reporting
Sustainable Supply Chain Management
Governance
Compliance
and Sustainability
Publication details
- Journal / source
- Mendeley Data
- Publication date
- 28 March 2026
- Language
- EN
- Document type
- Dataset