PG2Net
PG2Net:Personalized and Group Preference Guided Network for Next Place Prediction
Datasets
Experiment results on two Foursquare check-in datasets [NYC,TKY] and one mobile phone dataset [CDRs]
Requirements
- Python 3.6+
- Pytorch 1.4+
Project Structure
- baselines
- codes
- main.py # setting parameters
- train.py # train model
- model.py # define models
- utils.py # define tools
Usage
- The "distance.pkl" is generated by the "caculate_poi_distance" function in the "utils.py".
- The "cid_time.pkl" is generated by the "caculate_time_cid" function in the "utils.py".
- The "loc.emb" and "cid.emb" are generated by the "construc_graph" function in the "utils.py" and Node2vec.
- Train model #python main.py