ML-Approach to Qualimetry: GNNs in Value Assessment

Korostelev Sergey P

Keywords: integrated development of territories, cost, value, qualimetry, neural networks, stakeholders, Generative Neural Networks, ML-based approach

TL;DR: Sustainable territorial development needs multifaceted value assessment. Classical qualimetry is expert-heavy. An ML/GNN approach automates it. Tests confirm GNNs can build trees/calc weights. AI boosts practicality/speed.

Abstract: For sustainable Integrated Territorial Development (ITD), evaluating multifaceted value (social, ecological, etc.) is necessary, beyond traditional economic cost. Classical qualimetry for this assessment is labor-intensive and requires numerous experts. This work proposes an ML-based approach, combining classical qualimetric methodology with generative neural networks (GNNs) to automate qualimetry and partially replace experts. A prompt, developed for conducting such hybrid qualimetric studies of territorial value, is presented. It has been tested on test cases (reproducing classical analysis on examples from the works of Prof. Azgaldov G.G.). Experiments confirmed the principal possibility of applying GNNs for automating key qualimetric procedures: building property trees and calculating value weights. Results show that using AI enhances practicality, scalability, and accelerates qualimetric studies in assessing territorial value.

Primary Area: Evaluation (e.g., methodology, meta studies, replicability and validity, human-in-the-loop)

Flagged For Ethics Review: true

Submission Number: 16524