Skip to content
forked from skx300/DyHATR

[ECML-PKDD'20] Modeling Dynamic Heterogeneous Network for Link Prediction using Hierarchical Attention with Temporal RNN

License

Notifications You must be signed in to change notification settings

xuehansheng/DyHATR

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

DyHATR

This is the source code for ECML-PKDD 2020 paper "Modeling Dynamic Heterogeneous Network for Link Prediction using Hierarchical Attention with Temporal RNN".

Overview

image The novel dynamic heterogeneous network embedding method, termed as DyHATR, which uses hierarchical attention (HAT) to learn heterogeneous information and incorporates recurrent neural networks with temporal attention (Temporal Attentive RNN) to capture evolutionary patterns.

Requirement

Python 3.6
networkx == 1.11
numpy == 1.18
sklearn == 0.22
tensorflow == 1.14

Dataset

In this resposity, we provide EComm dataset as an example, you can also download all the other datasets from the SNAP platform (Twitter, and Math-Overflow). Besides, you can also use your own Dynamic Heterogeneous Networks dateset, as long as it fits the following template.

Node_one	Node_two	Edge_type	Timestamp
  n1		  n2		  e1		   1
  n1		  n2		  e2		   2
  n1		  n3		  e2		   2
  .
  .

Example Usage

To reproduce the experiments on EComm dataset, simply run:

python3 src/main.py

Reference

All readers are welcome to star/fork this repository and use it to reproduce our experiments or train your own data. Please kindly cite our paper:

@inproceedings{Xue2020DyHATR,
  title     = {Modeling Dynamic Heterogeneous Network forLink Prediction using Hierarchical Attentionwith Temporal RNN},
  author    = {Xue, Hansheng and Yang, Luwei and Jiang, Wen and Wei, Yi and Hu, Yi and Lin, Yu},
  booktitle = {ECML-PKDD},
  year      = {2020}
}

Acknowledgement

The original version of this code base was originally forked from GraphSAGE, and GAT, and we owe many thanks to these authors for making their code available. If you have some questions about the code or paper, you are welcome to open an issue or send us an email. We will respond to that as soon as possible.

About

[ECML-PKDD'20] Modeling Dynamic Heterogeneous Network for Link Prediction using Hierarchical Attention with Temporal RNN

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%