AI-enhanced MCDM framework for quantifying policy convergence and stake holder alignment in African mineral resource governance
# SCIPRA: Stakeholder-Centric Investment–Regulatory Policy Architecture
This repository contains the reproducibility package for the manuscript:
**Bridging the Investment-Regulatory-Stakeholder Divide: An AI-Enhanced MCDM Framework for Mineral Resource Policy Convergence in Africa**
The repository supports transparent replication of the SCIPRA framework, including the mathematical proofs, NLP–SVM pipeline, stakeholder salience scoring, and supplementary figures.
---
## Repository structure
```text
SCIPRA_GitHub_Package/
├── appendices/ # Supplementary materials and reference files
├── code/ # Reproducible Python scripts
├── data/
│ ├── raw/ # Public-source raw data placeholders
│ └── processed/ # Processed/derived datasets
├── docs/ # Data access and reproducibility notes
├── figures/
│ ├── svg/ # Camera-ready SVG figures
│ └── tiff_placeholder/# Placeholder for journal TIFF exports
├── results/ # Output files generated by scripts
├── CITATION.cff
├── LICENSE
├── README.md
└── requirements.txt
```
---
## Supplementary material map
- **Appendix A**: Mathematical proofs and formal derivations for PCI/RPCI boundedness, monotone convergence, and normalisation.
- **Appendix B**: NLP–SVM pipeline specification, lexicons, corpus metadata, and open-source data access pathway.
- **Appendix C**: Explicit derivation of stakeholder salience attributes: Power (P), Legitimacy (L), and Urgency (U).
---
## Reproducibility workflow
1. Install dependencies:
```bash
pip install -r requirements.txt
```
2. Prepare the public-source corpus using the access pathway described in `docs/DATA_ACCESS.md`.
3. Run the full end-to-end analysis (NLP-SVM + PCI/RPCI):
```bash
python code/execute_full_analysis.py
```
4. Generate the canonical manuscript table values:
```bash
python code/generate_tables.py
```
This produces the authoritative numbers for Tables 5 and 7. The manuscript reports the …