Research map / Graph learning / Knowledge and temporal graphs
UUKG: Unified Urban Knowledge Graph Dataset for Urban Spatiotemporal Prediction
Yansong Ning, Hao Liu, Hao Henry Wang, Zhenyu Zeng, Hui Xiong
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use non-Euclidean (hyperbolic/spherical) instead of Euclidean KG embeddings (ablation) | UrbanKG embedding · NYC link prediction | MRR: .516 -> .545 | Improved |
| use product-space embedding (GIE) instead of single hyperbolic space (best MuRP) (ablation) | UrbanKG embedding · NYC link prediction | MRR: .545 -> .573 | Improved |
| use product-space embedding (GIE) instead of single hyperbolic space (best MuRP) (ablation) | UrbanKG embedding · CHI link prediction | MRR: .519 -> .552 | Improved |
| use complex-space embeddings (RotatE, ComplEx, QuatE) instead of other Euclidean models (ablation) | UrbanKG embedding · NYC link prediction | MRR: .507 -> .274 | Hurt |
| concatenate product-space UrbanKG embeddings to STGCN input | STGCN · NYC Taxi | MAE/RMSE: 0.835/2.069 -> 0.781/1.942 | Improved |
| concatenate product-space UrbanKG embeddings to MTGNN input | MTGNN · NYC Taxi | MAE/RMSE: 1.289/2.275 -> 1.183/2.118 | Improved |
| concatenate product-space UrbanKG embeddings to AGCRN input | AGCRN · NYC Crime | Micro/Macro-F1: 65.98/60.62 -> 67.75/61.88 | Improved |
| concatenate product-space UrbanKG embeddings to HGCN input | HGCN · CHI Mobility | MAE/RMSE: 0.661/1.277 -> 0.637/1.146 | Improved |
| add UrbanKG embeddings from any space (vs none) (ablation) | USTP models · NYC Taxi and Crime | MAE / F1 | Improved |
| use product-space embedding instead of Euclidean/hyperbolic/spherical embedding for downstream USTP (ablation) | USTP models · NYC Taxi and Crime | MAE / F1 | Improved |
| increase product embedding dimension beyond 32 (ablation) | ASTGCN with UrbanKG embedding · NYC Taxi and Crime | MAE / F1 | No clear effect |
| raise embedding dimension from 32 to 150 for link prediction (ablation) | GIE · NYC link prediction | MRR: .573 -> .587 | Improved |
Closest papers by meaning
- UrbanKGent: A Unified Large Language Model Agent Framework for Urban Knowledge Graph Construction (NeurIPS 2024 · similarity 0.88)
- UrbanDataLayer: A Unified Data Pipeline for Urban Science (NeurIPS 2024 · similarity 0.86)
- Get Rid of Isolation: A Continuous Multi-task Spatio-Temporal Learning Framework (NeurIPS 2024 · similarity 0.86)
- Diffusion Transformers as Open-World Spatiotemporal Foundation Models (NeurIPS 2025 · similarity 0.85)
- Cross-City Latent Space Alignment for Consistency Region Embedding (ICML 2025 · similarity 0.85)
- GeoPro-Net: Learning Interpretable Spatiotemporal Prediction Models Through Statistically-Guided Geo-Prototyping (AAAI 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
