Researchers have introduced UrbanDS, a graph-guided large language model (LLM) multi-agent system designed to automate data-intensive urban tasks, according to a paper published on arxiv.org.
The system addresses challenges in urban data scenarios, which involve “large-scale and multi-sourced” data with “complex spatial, temporal, and semantic relationships,” according to the paper. UrbanDS constructs a unified dataset graph to organize reusable dataset skills and relationships. A Data Profiling Agent creates a skill for each dataset, while a Relation Agent identifies and integrates dataset relationships into the graph.
At runtime, a Planner Agent retrieves relevant datasets from the graph and generates execution plans. Multiple Execution Agents perform data processing and analysis, sharing their progress through common memory. A Report Agent then synthesizes experimental logs into a report that can be refined based on user feedback, according to arxiv.org.
The researchers also developed UrbanDS-Bench, “an urban data science benchmark covering representative data analysis and modeling tasks.” Experiments on both general and urban benchmarks showed that “UrbanDS consistently outperforms existing data science agents on data-intensive tasks,” the paper states.
According to arxiv.org, the system has been deployed on the urban operations platform of Dongxihu District in Wuhan, China, “demonstrating its effectiveness in real-world urban applications.”