Codes for WWW'19 Paper-DPLink: User Identity Linkage via Deep Neural Network From Heterogeneous Mobility Data
PyTorch implementation for DPLink: User Identity Linkage via Deep Neural Network From Heterogeneous Mobility Data. Jie Feng, Mingyang Zhang, Huandong Wang, Zeyu Yang, Chao Zhang, Yong Li, Depeng Jin. WWW 2019. If you find our code is useful for your research, you can cite our paper by:
@inproceedings{feng2019dplink,
title={DPLink: User Identity Linkage via Deep Neural Network From Heterogeneous Mobility Data},
author={Feng, Jie and Zhang, Mingyang and Wang, Huandong and Yang, Zeyu and Zhang, Chao and Li, Yong and Jin, Depeng},
booktitle={The World Wide Web Conference},
pages={459--469},
year={2019},
organization={ACM}
}
To train a new model (default settings are recorded in the run.py)
python run.py --data=foursquare --model=ERPC --pretrain=1 --pretrain_unit=ERCF
E: embedding, R: rnn, P: pooling, C: co-attention, F: fully connected network. ERPC is the default model in paper, model name can also be ERC(without pooling). ERCF is the default pretrain mode in paper, which means all the components in the model are pretrained. You can choose E, R, C, F for only pretrain selected component and N is for non-pretrain.
Baselines from traditional baselines, TULER and t2vec. Some codes from OpenNMT-py, InferSent and awd-lstm-lm.