Capturing Context-Aware Route Choice Semantics for Trajectory Representation Learning

Overview

This is the official implementation of paper "Capturing Context-Aware Route Choice Semantics for Trajectory Representation Learning" [arXiv], which learns trajectory embeddings by incorporating context-aware route choice semantics as an inductive bias.

Introduction

Trajectory representation learning (TRL) aims to encode raw trajectory data into low-dimensional embeddings for downstream tasks such as travel time estimation, mobility prediction, and trajectory similarity analysis. From a behavioral perspective, a trajectory reflects a sequence of route choices within an urban environment. However, most existing TRL methods ignore this underlying decision-making process and instead treat trajectories as static, passive spatiotemporal sequences, thereby limiting the semantic richness of the learned representations. To bridge this gap, we propose CORE, a TRL framework that integrates context-aware route choice semantics into trajectory embeddings. CORE first incorporates a multi-granular Environment Perception Module, which leverages large language models (LLMs) to distill environmental semantics from point of interest (POI) distributions, thereby constructing a context-enriched road network. Building upon this backbone, CORE employs a Route Choice Encoder with a mixture-of-experts (MoE) architecture, which captures route choice patterns by jointly leveraging the context-enriched road network and navigational factors. Finally, a Transformer encoder aggregates the route-choice-aware representations into a global trajectory embedding. Extensive experiments on 4 real-world datasets across 6 downstream tasks demonstrate that CORE consistently outperforms 12 state-of-the-art TRL methods, achieving an average improvement of 9.79% over the best-performing baseline. The overview of our framework is shown as below:

Installation

Environment

• Tested OS: Linux
• Python >= 3.9

Dependencies

1. Install PyTorch with the correct CUDA version.
2. Use the pip install -r requirements.txt command to install all of the Python modules and packages used in this project.

Usage Instructions

1. Data Preparation

Please follow these steps to prepare your dataset:

1. Create a new directory for your dataset inside the data/ directory.

2. Within your new dataset directory, create two subdirectories: poi/ and traj/.

3. Place your data files (e.g., poi.csv, roadmap.geo, roadmap.rel, trajectory_*.csv) into the corresponding subdirectories. For details on the specific function and placement of each file, please refer to the example structure for the Beijing dataset in the Code Structure section above.

2. Data Preprocessing

cd data

python generate_road_freq.py --dataset <city_name>
python generate_road_network_info.py --dataset <city_name>
python generate_aug_traj_for_pretrain.py --dataset <city_name>
python generate_traj_sim_data.py --dataset <city_name>
python generate_path_rank_data.py --dataset <city_name>

3. LLM Description & Text Embedding

cd llm_script

python generate_road_poi_data.py --dataset <city_name>
python generate_road_poi_prompt.py --dataset <city_name>
python generate_road_poi_description.py --dataset <city_name>
python generate_road_poi_embedding.py --dataset <city_name>

python generate_grid_poi_data.py --dataset <city_name>
python generate_grid_poi_prompt.py --dataset <city_name>
python generate_grid_poi_description.py --dataset <city_name>
python generate_grid_poi_embedding.py --dataset <city_name>

4. Model Pre-Training

python pretrain.py --dataset <city_name>

5. Running Downstream Tasks

Road Label Prediction:

python train_road_label.py --dataset <city_name>

Trajectory Destination Prediction:

python train_des_pred.py --dataset <city_name>

Travel Time Estimation:

python train_tte.py --dataset <city_name>

Similar Trajectory Retrieval:

python run_traj_sim.py --dataset <city_name>

Path Ranking:

python train_path_rank.py --dataset <city_name>

Trajectory Generation:

Please refer to caoji2001/HOSER.