Docx

https://docs.google.com/document/d/1Ee2X4yjXX8eQaLQ8T6nr3z7CDE2E2EPODjCjAtdDCRI/edit?usp=sharing

Code Details

• Each python file contains the explaination that how that function/class are working
• parse.py , world.py , register.py and main.py are the trival files which contains basic info. like parameter used and locations
• Order to look the code is dataloader.py -> parse.py -> world.py -> register.py -> model.py -> utils.py -> procedure.py -> main.py

Update

• A new class is made called Loader2() for movielens100k dataset
• Since for moviesLens it has 610 users_id and about 9k distinct movie_id ranging from 0 to 190k and rating from 0 to 5. The data is preprocessed to make train.txt and test.txt and converted to user_id and movie_id interaction matrix. As the highest movie_id is 190K so it will make graph with 190K column so i have mapped the movie_id to reduce the final dimension from (610 x 190k) to (610 x 9k).

Things to remember

• while using gowalla dataset make sure to increase batchsize to 100 in parse.py otherwise it will take long time to run
• while using ml100k dataset in gowalla type loader class that is loader2 make sure to register it in register.py (already commented)
• while using ml100k in lastfm type class that ml100k() , again register in register.py
• use this command cd code && python main.py --decay=1e-4 --lr=0.0005 --layer=4 --seed=2020 --dataset="ml100k" --topks="[50]" --recdim=32 and change name of dataset accordingly under quotes
• if you are retraining the model for ml100k dataset make sure to delete "s_pre_adj_mat.npz" file in ml100k folder as it stores the matrix data when using gowalla type loader class.

Dataset

• MovieLens - https://grouplens.org/datasets/movielens/
• MindReader - https://mindreader.tech/dataset/

2020-09:

• Change the print format of each epoch
• Add Cpp Extension in code/sources/ for negative sampling. To use the extension, please install pybind11 and cppimport under your environment

---

LightGCN-pytorch

This is the Pytorch implementation for our SIGIR 2020 paper:

SIGIR 2020. Xiangnan He, Kuan Deng ,Xiang Wang, Yan Li, Yongdong Zhang, Meng Wang(2020). LightGCN: Simplifying and Powering Graph Convolution Network for Recommendation, Paper in arXiv.

Author: Prof. Xiangnan He (staff.ustc.edu.cn/~hexn/)

(Also see Tensorflow implementation)

Introduction

In this work, we aim to simplify the design of GCN to make it more concise and appropriate for recommendation. We propose a new model named LightGCN,including only the most essential component in GCN—neighborhood aggregation—for collaborative filtering

Enviroment Requirement

pip install -r requirements.txt

Dataset

We provide three processed datasets: Gowalla, Yelp2018 and Amazon-book and one small dataset LastFM.

see more in dataloader.py

An example to run a 3-layer LightGCN

run LightGCN on Gowalla dataset:

• change base directory

Change ROOT_PATH in code/world.py

• command

cd code && python main.py --decay=1e-4 --lr=0.001 --layer=3 --seed=2020 --dataset="gowalla" --topks="[20]" --recdim=64

• log output

...
======================
EPOCH[5/1000]
BPR[sample time][16.2=15.84+0.42]
[saved][[BPR[aver loss1.128e-01]]
[0;30;43m[TEST][0m
{'precision': array([0.03315359]), 'recall': array([0.10711388]), 'ndcg': array([0.08940792])}
[TOTAL TIME] 35.9975962638855
...
======================
EPOCH[116/1000]
BPR[sample time][16.9=16.60+0.45]
[saved][[BPR[aver loss2.056e-02]]
[TOTAL TIME] 30.99874997138977
...

NOTE:

1. Even though we offer the code to split user-item matrix for matrix multiplication, we strongly suggest you don't enable it since it will extremely slow down the training speed.
2. If you feel the test process is slow, try to increase the testbatch and enable multicore(Windows system may encounter problems with multicore option enabled)
3. Use tensorboard option, it's good.
4. Since we fix the seed(--seed=2020 ) of numpy and torch in the beginning, if you run the command as we do above, you should have the exact output log despite the running time (check your output of epoch 5 and epoch 116).

Extend:

• If you want to run lightGCN on your own dataset, you should go to dataloader.py, and implement a dataloader inherited from BasicDataset. Then register it in register.py.
• If you want to run your own models on the datasets we offer, you should go to model.py, and implement a model inherited from BasicModel. Then register it in register.py.
• If you want to run your own sampling methods on the datasets and models we offer, you should go to Procedure.py, and implement a function. Then modify the corresponding code in main.py

Results

all metrics is under top-20

pytorch version results (stop at 1000 epochs):

(for seed=2020)

• gowalla:

	Recall	ndcg	precision
layer=1	0.1687	0.1417	0.05106
layer=2	0.1786	0.1524	0.05456
layer=3	0.1824	0.1547	0.05589
layer=4	0.1825	0.1537	0.05576

• yelp2018

	Recall	ndcg	precision
layer=1	0.05604	0.04557	0.02519
layer=2	0.05988	0.04956	0.0271
layer=3	0.06347	0.05238	0.0285
layer=4	0.06515	0.05325	0.02917