Data, Code, and Audit Materials for "Coupling AI-Estimated Streetscape Impressions with Network Topology for Historic Urban Landscape Micro-Renewal Prioritization: Evidence from Nanluoguxiang, Beijing"

This repository contains the derived data, analysis code, prompts, model configurations, street-network files, reproducibility documentation, and quality-control audit materials supporting the manuscript “Coupling AI-Estimated Streetscape Impressions with Network Topology for Historic Urban Landscape Micro-Renewal Prioritization: Evidence from Nanluoguxiang, Beijing”. The repository includes the deduplicated directional-image index, panorama-level semantic indicators, Qwen3-VL-Plus scoring and adjudication records, final modeling data for 535 panoramas, spatially grouped cross-validation outputs, accumulated local effect results, ridge-regression coefficients, street-network topology data, stochastic multi-criteria priority results, bounded counterfactual scenario outputs, and validation scripts. Raw street-view image pixels are not redistributed because they are governed by the source platform’s terms of use. Reproducible panorama identifiers, coordinates, headings, imagery dates where available, derived semantic indicators, AI-estimated impression scores, and analytical outputs are provided. Qwen3-VL-Plus outputs should be interpreted as AI-estimated visual impressions rather than human-perception ground truth. Data and documentation are licensed under the Creative Commons Attribution 4.0 International license. Source code is licensed separately under the MIT License, as specified in the repository license files.