Open josix opened 4 years ago
Hi @josix,
please feel free to do so, I would highly appreciate it! Better knowledge graph support is something I would like to see supported in PyG :) In addition to KGAT, we should also add some of the datasets used in their experiments.
Hello @rusty1s , thanks for your reply. I'm so glad you also think so :) I will start to implement it from adding a new dataset that KGAT used in its experiment and the KG supported data representation. Do you have any suggested code examples that I can follow to avoid my codes are not totally consistent with yours? Or I could start to write first then we can discuss after I push my first draft PR?
Which dataset do you want to implement? I think you can just start sending a WIP pull request following the implementations if other datasets, and we can discuss the details as you go :) Thank you so much!
You may want to look at the AMiner
dataset for how to implement heterogeneous graphs.
Okay! Thanks for your advice! I will implement last.fm dataset these days including U-I Interaction part and KG part.
Okay! Thanks for your advice! I will implement last.fm dataset these days including U-I Interaction part and KG part.
@josix Hi! Have you finished?
Okay! Thanks for your advice! I will implement last.fm dataset these days including U-I Interaction part and KG part.
Hey, any updates?
Maybe, this could be useful: KGAT implementation in PyTorch: https://github.com/LunaBlack/KGAT-pytorch
🚀 Feature
Hello @rusty1s,
Thanks for your work on this project. This is an awesome package. I am looking for a more efficient implemented KGAT models, which is a published recommendation method on KDD19’; however it seems that it isn't not implemented in the examples. Could I try to make a PR and add this model to the examples?
Motivation
I guess it would be better if there is a PyG version of KGAT so that other people can refer to it and implement some similar models
Additional context
paper: https://xiangwang1223.github.io/papers/KGAT_final.pdf code: https://github.com/xiangwang1223/knowledge_graph_attention_network