This is the implementation of the manuscript entitled: STRelay: A Universal Spatio-Temporal Relaying Framework for Location Prediction over Human Trajectory Data

We upload the STRelay with Graph-Flashback (the best-performing base model) for your reference.

DATA and Pretrained Graphs

The datasets are pretrained graphs available here: https://drive.google.com/drive/folders/1WJOfoMPUj3nVJOqJKwPtDS-N8G8-pI6s?usp=drive_link

Please download from the link above, and place the two folders data and POI_graph in the current directory.

Requirements

python: 3.8
torch: 2.3.0
numpy
tqdm
scipy

To reproduce STRelay with best performing base model in our paper, please run following scripts:

python -u train.py --dataset Istanbul.txt --trans_loc_file POI_graph/Istanbul_scheme2_transe_loc_temporal_20.pkl --trans_interact_file POI_graph/Istanbul_scheme2_transe_user-loc_100.pkl  --gpu 0  --batch-size 128 --STRelay True --log_file results/strelay_istanbul

python -u train.py --dataset Tokyo.txt --trans_loc_file POI_graph/Tokyo_scheme2_transe_loc_temporal_20.pkl --trans_interact_file POI_graph/Tokyo_scheme2_transe_user-loc_100.pkl --gpu 0 --batch-size 64 --STRelay True --log_file results/strelay_tokyo

python -u train.py --dataset Singapore.txt --trans_loc_file POI_graph/Singapore_scheme2_transe_loc_temporal_20.pkl --trans_interact_file POI_graph/Singapore_scheme2_transe_user-loc_100.pkl  --gpu 0  --batch-size 64 --STRelay True --log_file results/strelay_singapore 

python -u train.py --dataset Moscow.txt --trans_loc_file POI_graph/Moscow_scheme2_transe_loc_temporal_20.pkl --trans_interact_file POI_graph/Moscow_scheme2_transe_user-loc_100.pkl --gpu 0 --batch-size 128 --STRelay True --log_file results/strelay_moscow