ebay / kprn Goto Github PK

Reasoning Over Knowledge Graph Paths for Recommendation

License: Other

Python 36.27% Lua 53.56% Shell 8.68% Jupyter Notebook 1.49%

kprn's Introduction

Reasoning Over Knowledge Graph Paths for Recommendation

This is code related to the AAAI 2019 paper "Explainable Reasoning over Knowledge Graphs for Recommendation.". The code makes extensive use of machine learning techniques, and will be useful for training and prediction of recommendation attributes of media, or other items as described in the paper.

Platform Requirements

This code requires Python(2.7 or 3.5) and Lua(5.3). Please ensure the runtime environments for these are installed. The details could be found in the readMe.pdf.

Steps to Build a Model File in Training Model & Steps to Make Predictions

The model details could be found through readMe.pdf.

Attribution and Acknowledgements

Acknowledgement and thanks to others for open source work used in this project. Code used in this project is available from the following sources.

https://github.com/rajarshd/ChainsofReasoning
Author: Rajarshi Das
See Chains of Reasoning over Entities, Relations, and Text using Recurrent Neural Networks
Licensed under at least Section D5 of Github Terms of Service..
https://github.com/hexiangnan/neural_collaborative_filtering
Author: Dr. Xiangnan He (http://www.comp.nus.edu.sg/~xiangnan/)
See Xiangnan He, Lizi Liao, Hanwang Zhang, Liqiang Nie, Xia Hu and Tat-Seng Chua (2017). Neural Collaborative Filtering. In Proceedings of WWW '17, Perth, Australia, April 03-07, 2017.
Licensed under Apache 2.0.
https://github.com/hexiangnan/neural_factorization_machine
Author: Dr. Xiangnan He (http://www.comp.nus.edu.sg/~xiangnan/)
See Xiangnan He and Tat-Seng Chua (2017). Neural Factorization Machines for Sparse Predictive Analytics. In Proceedings of SIGIR '17, Shinjuku, Tokyo, Japan, August 07-11, 2017.
Licensed under at least Section D5 of Github Terms of Service.
https://github.com/HKUST-KnowComp/FMG
See Meta-Graph Based Recommendation Fusion over Heterogeneous Information Networks
Licensed under at least Section D5 of Github Terms of Service.

License

Modifications Copyright 2018 eBay Inc.
Authors/Developers of Modifications: Dingxian, Wang ([email protected]) and Canran, Xu ([email protected])
New code and modifications to code are licensed under the MIT License.. See LICENSE for the license text.

kprn's People

Contributors

Stargazers

Watchers

Forkers

platakwon zscdumin gegetang wurentidai ggsonic jianzfb notfoundgg aliciaframe shubhampachori12110095 marsxdm dario-github csc19960608 365318663 uctoronto ethanlee555 leiloong 67in growerwang chenjie04 mmliang scorpjd farisology limingmingli321 tanxiangyueer sunyatong regzhuce xcgfth floatsdsds jessinra linessiex li-study todun wangxuekui tlwzzy zscwind tiansheng111 xiaoqingwang shengyupei jufangshen swan815 m0llydayu ccfbupt hawksilent hongwang-sun semsevens robstens cindychen1995 mauriziofd liu429536796 quanjinwusheng chicke-ai xjliu1204 rcdnn shiyinw liuweiping2020 eminemrain ywzhang dpavot yueyedeai juneqiong daguo97 haeuuu yuanyuansiyuan doctorjia phac123 gabriel1521 isabella232 gl123456 zyy598 liangzongchang tpnguyen sunatthegilddotcom jinyu9 wangjiannan98 njust-csc jaemyoung pushpull7381 crystal22 hbamboo stjordanis qpc-github yixiangxi

kprn's Issues

readme中提到的“Path RNN”就是论文里的KPRN模型吗？

can you complete the code and file that can make the model go through?

TEM

Can I get TEM code ？

About embedding

Hi,

Sorry to disturb, I would like to know which method you used to embed the entities and relations in a certain path.Thanks.

Is the shared code file incomplete?

I couldn't find the script files.
A
path_config.sh: config file
run_path_find.sh: run run_path.sh
C
data.sh
movie_run_exp.py
run_test.sh
D
python ItemPop.py
train_nfm.sh

How to get all data?

I can't even get prepare data,like music :

song_person.dict: key: song id,value: a list of all the related song person id
person_song.dict: reverse of the song_person.dict
song_type.dict: key: song id,value: and a list of all the related song type id
type_song.dict: key: reverse of the song_type.dict

above is just a little.I want to Repete your experiment,so can you help to get all the data.I have read your 'readMe.pdf'.But I can't get data from the url you suggest.
My email [email protected],if you can help me,I should fell grateful.Thank you!

Questions about path extraction

Hi, I have one question regarding the path extraction. Do you use all the paths within length 6 between an user-item pair or random sample some of them. Because there maybe thousands of paths bewteen a pair in Movielens Dataset.
Thanks for your attention :)

Question about add_relation_label

I'm confused by the function "get_relation" defined in "add_relation_label.py". It appears that the authors assume that the song is represented by "s", the user is represented by "u", the artist is represented by "p" and the type of song is represented by "t". However, the song id and user id token directly from the original KG data is something like "++5wYjoMgQHoRuD3GbbvmphZbBBwymzv5Q4l8sywtuU=" or "NYE7sp/xNTq3AMFmhLT9kHdtqItEOA34m7kbGUr4Y0k=" which is quite different from the assumption above. What am I supposed to do? Reindex each user and song? I didn't find any instructions about this problem in "readMe.pdf". Please, help me!

Where does the Knowledge Graphs (KG) they used come from? Do they are constructed by authors?

Sorry, I cannot find where introduce to the KGs. Do the authors use the public KGs? What does the KG in Table 1 means?

How to preprocess the data?

I don't know how to generate all the dictions. For example, 'song_person.dict', what should i put in this file? I've read 'read me.pdf', but i can't find the answer. Thanks for your help.

ebay / kprn Goto Github PK

kprn's Introduction

Reasoning Over Knowledge Graph Paths for Recommendation

Platform Requirements

Steps to Build a Model File in Training Model & Steps to Make Predictions

Attribution and Acknowledgements

License

kprn's People

Contributors

Stargazers

Watchers

Forkers

kprn's Issues

Recommend Projects

Recommend Topics

Recommend Org