Supplementary materials: AI-augmented urban indicator development for the Ghana Urban Observatory — evaluation code and results data

Supplementary evaluation code and results data accompanying the manuscript: Zingbagba, M. (under review). AI-augmented urban indicator development for the Ghana Urban Observatory: A semantic search, justification, and gap analysis system aligned to SDG 11. Journal of Urban Technology.This deposit contains seven files:(1) guo_figures.py — Python script reproducing all three publication figures. Requires: pip install matplotlib numpy. No API credentials required.(2) guo_baseline_comparison.py — BM25 vs. semantic search evaluation script across 20 queries spanning all 12 GUO indicator domains. Computes Precision@5 and Mean Reciprocal Rank. Requires a live Supabase instance and OpenAI API key (see README).(3) guo_justification_assessment.py — Interactive quality assessment rubric for AI-generated indicator justification documents. Scores 10 sampled indicators across three criteria on a three-point scale. Requires a live Supabase instance (see README).(4) guo_interrater_agreement.py — Cohen's Kappa and percentage agreement calculator for use when two evaluators complete the quality assessment independently. No credentials required.(5) baseline_comparison_table.csv — Per-query and aggregate results of the BM25 vs. semantic search evaluation (20 queries, Precision@5 and MRR).(6) justification_scores_evaluator.csv — Quality assessment scores for 10 sampled indicators across three criteria (policy alignment accuracy, Ghana contextual relevance, alternative indicators reasoning).(7) README.md — Full documentation of the system architecture, file contents, reproduction instructions, and data access contact.Full implementation code (main.py, guo_ingest.py) and the 94-indicator metadata corpus are available on request from the corresponding author: mark.zingbagba@expertisefrance.fr