Graph Neural Networks for City Policy Recommendations as a Link Prediction Task

Abstract

Graph Neural Networks (GNNs) have become a widely utilized tool in recommender systems in various contexts. While recommendation tasks can be approached using a multitude of data structures and types, graph-structured data is particularly well-suited for this domain, as graphs naturally capture a variety of relationships and interactions between entities. By leveraging graph representation learning, we can effectively encode these complex dependencies, enabling robust and context-aware recommendations. We use this methodology in the domain of policy recommendations for urban centers. To recommend policies, we would learn the complex local and global relationships between cities, their environmental features, and currently implemented policies. We construct a graph structure relating cities, implemented policies, and city features, and formulate the policy recommendation task as a GNN link prediction problem, demonstrating its potential to scale data-driven urban governance.