Skip to content

xiangwang1223/knowledge_graph_attention_network

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

60 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Knowledge Graph Attention Network

This is our Tensorflow implementation for the paper:

Xiang Wang, Xiangnan He, Yixin Cao, Meng Liu and Tat-Seng Chua (2019). KGAT: Knowledge Graph Attention Network for Recommendation. Paper in ACM DL or Paper in arXiv. In KDD'19, Anchorage, Alaska, USA, August 4-8, 2019.

Author: Dr. Xiang Wang (xiangwang at u.nus.edu)

Introduction

Knowledge Graph Attention Network (KGAT) is a new recommendation framework tailored to knowledge-aware personalized recommendation. Built upon the graph neural network framework, KGAT explicitly models the high-order relations in collaborative knowledge graph to provide better recommendation with item side information.

Citation

If you want to use our codes and datasets in your research, please cite:

@inproceedings{KGAT19,
  author    = {Xiang Wang and
               Xiangnan He and
               Yixin Cao and
               Meng Liu and
               Tat{-}Seng Chua},
  title     = {{KGAT:} Knowledge Graph Attention Network for Recommendation},
  booktitle = {{KDD}},
  pages     = {950--958},
  year      = {2019}
}

Environment Requirement

The code has been tested running under Python 3.6.5. The required packages are as follows:

  • tensorflow == 1.12.0
  • numpy == 1.15.4
  • scipy == 1.1.0
  • sklearn == 0.20.0

Reproducibility & Example to Run the Codes

To demonstrate the reproducibility of the best performance reported in our paper and faciliate researchers to track whether the model status is consistent with ours, we provide the best parameter settings (might be different for the custormized datasets) in the scripts, and provide the log for our trainings.

The instruction of commands has been clearly stated in the codes (see the parser function in Model/utility/parser.py).

  • Yelp2018 dataset
python Main.py --model_type kgat --alg_type bi --dataset yelp2018 --regs [1e-5,1e-5] --layer_size [64,32,16] --embed_size 64 --lr 0.0001 --epoch 1000 --verbose 50 --save_flag 1 --pretrain -1 --batch_size 1024 --node_dropout [0.1] --mess_dropout [0.1,0.1,0.1] --use_att True --use_kge True
  • Amazon-book dataset
python Main.py --model_type kgat --alg_type bi --dataset amazon-book --regs [1e-5,1e-5] --layer_size [64,32,16] --embed_size 64 --lr 0.0001 --epoch 1000 --verbose 50 --save_flag 1 --pretrain -1 --batch_size 1024 --node_dropout [0.1] --mess_dropout [0.1,0.1,0.1] --use_att True --use_kge True
  • Last-fm dataset
python Main.py --model_type kgat --alg_type bi --dataset last-fm --regs [1e-5,1e-5] --layer_size [64,32,16] --embed_size 64 --lr 0.0001 --epoch 1000 --verbose 50 --save_flag 1 --pretrain -1 --batch_size 1024 --node_dropout [0.1] --mess_dropout [0.1,0.1,0.1] --use_att True --use_kge True

Some important arguments:

Dataset

We provide three processed datasets: Amazon-book, Last-FM, and Yelp2018.

  • You can find the full version of recommendation datasets via Amazon-book, Last-FM, and Yelp2018.
  • We follow KB4Rec to preprocess Amazon-book and Last-FM datasets, mapping items into Freebase entities via title matching if there is a mapping available.
Amazon-book Last-FM Yelp2018
User-Item Interaction #Users 70,679 23,566 45,919
#Items 24,915 48,123 45,538
#Interactions 847,733 3,034,796 1,185,068
Knowledge Graph #Entities 88,572 58,266 90,961
#Relations 39 9 42
#Triplets 2,557,746 464,567 1,853,704
  • train.txt

    • Train file.
    • Each line is a user with her/his positive interactions with items: (userID and a list of itemID).
  • test.txt

    • Test file (positive instances).
    • Each line is a user with her/his positive interactions with items: (userID and a list of itemID).
    • Note that here we treat all unobserved interactions as the negative instances when reporting performance.
  • user_list.txt

    • User file.
    • Each line is a triplet (org_id, remap_id) for one user, where org_id and remap_id represent the ID of such user in the original and our datasets, respectively.
  • item_list.txt

    • Item file.
    • Each line is a triplet (org_id, remap_id, freebase_id) for one item, where org_id, remap_id, and freebase_id represent the ID of such item in the original, our datasets, and freebase, respectively.
  • entity_list.txt

    • Entity file.
    • Each line is a triplet (freebase_id, remap_id) for one entity in knowledge graph, where freebase_id and remap_id represent the ID of such entity in freebase and our datasets, respectively.
  • relation_list.txt

    • Relation file.
    • Each line is a triplet (freebase_id, remap_id) for one relation in knowledge graph, where freebase_id and remap_id represent the ID of such relation in freebase and our datasets, respectively.

Acknowledgement

Any scientific publications that use our datasets should cite the following paper as the reference:

@inproceedings{KGAT19,
  author    = {Xiang Wang and
               Xiangnan He and
               Yixin Cao and
               Meng Liu and
               Tat-Seng Chua},
  title     = {KGAT: Knowledge Graph Attention Network for Recommendation},
  booktitle = {{KDD}},
  year      = {2019}
}

Nobody guarantees the correctness of the data, its suitability for any particular purpose, or the validity of results based on the use of the data set. The data set may be used for any research purposes under the following conditions:

  • The user must acknowledge the use of the data set in publications resulting from the use of the data set.
  • The user may not redistribute the data without separate permission.
  • The user may not try to deanonymise the data.
  • The user may not use this information for any commercial or revenue-bearing purposes without first obtaining permission from us.

Funding Source Acknowledgement

This research is supported by the National Research Foundation, Singapore under its International Research Centres in Singapore Funding Initiative. Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not reflect the views of National Research Foundation, Singapore.