README.md

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
  • Markov.py
  • DeepMove.py Paper
  • PLSPL.py Paper
  • LSTPM.py Paper
  • STAN.py paper
  • GETNext.py paper
• 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