English

BERT4Traj: Transformer Based Trajectory Reconstruction for Sparse Mobility Data

Machine Learning 2025-07-08 v1 Artificial Intelligence

Abstract

Understanding human mobility is essential for applications in public health, transportation, and urban planning. However, mobility data often suffers from sparsity due to limitations in data collection methods, such as infrequent GPS sampling or call detail record (CDR) data that only capture locations during communication events. To address this challenge, we propose BERT4Traj, a transformer based model that reconstructs complete mobility trajectories by predicting hidden visits in sparse movement sequences. Inspired by BERT's masked language modeling objective and self_attention mechanisms, BERT4Traj leverages spatial embeddings, temporal embeddings, and contextual background features such as demographics and anchor points. We evaluate BERT4Traj on real world CDR and GPS datasets collected in Kampala, Uganda, demonstrating that our approach significantly outperforms traditional models such as Markov Chains, KNN, RNNs, and LSTMs. Our results show that BERT4Traj effectively reconstructs detailed and continuous mobility trajectories, enhancing insights into human movement patterns.

Keywords

Cite

@article{arxiv.2507.03062,
  title  = {BERT4Traj: Transformer Based Trajectory Reconstruction for Sparse Mobility Data},
  author = {Hao Yang and Angela Yao and Christopher Whalen and Gengchen Mai},
  journal= {arXiv preprint arXiv:2507.03062},
  year   = {2025}
}

Comments

This paper was accepted at GIScience 2025

R2 v1 2026-07-01T03:45:47.489Z