TrajPrism introduces a multi-task benchmark with 300K real-world urban trajectories and 2.1M language-grounded task instances across three cities, plus proof-of-concept models showing large gaps versus geometry-only baselines.
Title resolution pending
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.AI 2years
2026 2verdicts
UNVERDICTED 2roles
background 1polarities
background 1representative citing papers
Presents an end-to-end system using LLM agents to add behavioral anomalies to simulated trajectories, then applies map routing and noise to generate realistic annotated anomaly datasets for mobility research.
citing papers explorer
-
TrajPrism: A Multi-Task Benchmark for Language-Grounded Urban Trajectory Understanding
TrajPrism introduces a multi-task benchmark with 300K real-world urban trajectories and 2.1M language-grounded task instances across three cities, plus proof-of-concept models showing large gaps versus geometry-only baselines.
-
Mobility Anomaly Generation using LLM-Driven Behavior with Kinematic Constraints
Presents an end-to-end system using LLM agents to add behavioral anomalies to simulated trajectories, then applies map routing and noise to generate realistic annotated anomaly datasets for mobility research.