July 16, 2026
UrbanTrace Paper Accepted to IEEE VIS 2026
News2026-07-16
The paper 'UrbanTrace: LLM-Assisted Discovery and Semantics-Aware Integration of Spatial Data' was accepted to IEEE VIS 2026.
The paper "UrbanTrace: LLM-Assisted Discovery and Semantics-Aware Integration of Spatial Data," authored by Sonia Castelo Quispe, Eden Wu, and João Rulff, Harish Doraiswamy (Microsoft Research India) and Juliana Freire, has been accepted to IEEE VIS 2026.
The paper tackles a pervasive problem in urban analytics: integrating heterogeneous spatial data is full of semantic traps, and GIS tools can't warn you when aggregating across mismatched boundaries silently corrupts an analysis. UrbanTrace turns this manual wrangling into a transparent, node-based workflow with context-aware AI agents, grounding LLMs in real data distributions so they can retrieve datasets from high-level goals and enforce valid spatial aggregations, while interactive views show how conclusions shift across spatial configurations.
The paper tackles a pervasive problem in urban analytics: integrating heterogeneous spatial data is full of semantic traps, and GIS tools can't warn you when aggregating across mismatched boundaries silently corrupts an analysis. UrbanTrace turns this manual wrangling into a transparent, node-based workflow with context-aware AI agents, grounding LLMs in real data distributions so they can retrieve datasets from high-level goals and enforce valid spatial aggregations, while interactive views show how conclusions shift across spatial configurations.