PPR-master

The implementation of the PPR model(Personalized POI Recommendation).

Paper Personalized POI Recommendation:Spatio-Temporal Representation Learning with Social Tie was accepted by DASFAA 2021.

Usage:

Some questions about Embedding and line.cpp, please refer to LINE (link: https://github.com/tangjianpku/LINE).

Install dependencies

pip install -r requirements.txt

Installation

Clone this repo.

git clone https://github.com/dsj96/PPR-master
cd PPR-master
chmod u+x train_PPR.sh

Function

1. gen_graph.py file is used for heterogeneous graph construction. Parameter theta is $\theta$ in Equ.2, and epsilon is $\varepsilon$ in Equ.6.
2. reconstruct.cpp file is used for densifying graph. Parameter -threshold is $\rho$.
3. line.cpp file is used for learning latent representations. Parameter -size is embedding dim $d$.
4. train.py file is used for training and evaluating the spatio-temporal neural network. Parameter DELT_T is the time constraint $\tau$, and INPUT_SIZE/2 is the embedding dim $d$. You could also change the HIDDEN_SIZE, EPOCH, LR, LAYERS OR TEST_SAMPLE_NUM.

Data

In our experiments, the Foursquare datasets are from https://sites.google.com/site/dbhongzhi/ (update: https://sites.google.com/view/hongzhi-yin/datasets). And the Gowalla and Brightkite dataset are from https://snap.stanford.edu/data/loc-gowalla.html and http://snap.stanford.edu/data/loc-Gowalla.html.

Data Format

We utilize the first 80% chronological check-ins of each user as the training set, the remaining 20% as the test data.

train_checkin_file.txt and test_checkin_file.txt : <USER ID> \t <CHECKIN TIME> \t <POI ID> \t <LONGITUDE> \t <LATITUDE>

friendship_file.txt : <USER ID>,<USER ID>

Training

You can train and evaluate the model by: ./train_PPR.sh

Or you can run the specific program file separately, but the parameters should be reasonable (Although we have set some default parameters).

eg: 
python3 gen_graph.py --input_path dataset/toyset/ --epsilon 0.5 --theta 24.0
python3 train.py --input_path dataset/toyset/ --input_size 16 --hidden_size 16 --layers 2 --lr 0.001 --delt_t 6.0 --epochs 20 --dr 0.2 --seed 1 --test_sample_num 300

Cite

Please cite our paper if you find this code useful for your research:

@inproceedings{dai2021personalized,
  title={Personalized POI Recommendation: Spatio-Temporal Representation Learning with Social Tie.},
  author={Dai, Shaojie and Yu, Yanwei and Fan, Hao and Dong, Junyu},
  booktitle={DASFAA (1)},
  pages={558--574},
  year={2021}
}