MMGCN: Multi-modal Graph Convolution Network for Personalized Recommendation of Micro-video

A novel multi-modal recommendation framework based on graph convolutional networks, explicitly modeling modal-specific user preferences to enhance micro-video recommendation.

Authors

Yinwei Wei¹, Xiang Wang², Liqiang Nie¹*, Xiangnan He³, Richang Hong⁴, Tat-Seng Chua²

¹ Shandong University, China
² National University of Singapore, Singapore
³ University of Science and Technology of China, China
⁴ Hefei University of Technology, China
* Corresponding author

Table of Contents

• Updates
• Introduction
• Method / Framework
• Installation
• Dataset / Benchmark
• Usage
• License

---

Links

- Paper: ACM MM'19
- Code Repository: [`GitHub`](https://github.com/iLearn-Lab/MM19-MMGCN)

---

Updates

- [10/2019] Paper presented at ACM MM'19.
- [01/2020] Initial release of the PyTorch implementation and toy datasets.

---

Introduction

This is the official PyTorch implementation for the paper MMGCN: Multi-modal Graph Convolution Network for Personalized Recommendation of Micro-video.

Multi-modal Graph Convolution Network is a novel multi-modal recommendation framework based on graph convolutional networks. It explicitly models modal-specific user preferences to enhance micro-video recommendation. In this repository, we provide the updated code and utilize a full-ranking strategy for both validation and testing.

---

Method / Framework

---

Installation

1. Clone the repository

git clone \[https://github.com/iLearn-Lab/MM19-MMGCN.git\](https://github.com/iLearn-Lab/MM19-MMGCN.git) cd MM19-MMGCN

2. Environment Requirements

The code has been tested running under Python 3.5.2. The required packages are as follows:

• Pytorch == 1.1.0
• torch-cluster == 1.4.2
• torch-geometric == 1.2.1
• torch-scatter == 1.2.0
• torch-sparse == 0.4.0
• numpy == 1.16.0

Install the dependencies using pip:

pip install torch==1.1.0 torchvision pip install torch-scatter==1.2.0 torch-sparse==0.4.0 torch-cluster==1.4.2 torch-geometric==1.2.1 pip install numpy==1.16.0

---

Dataset / Benchmark

We provide three processed datasets: Kwai, Tiktok, and Movielens.

Due to copyright restrictions, we cannot release the full datasets directly. You can find the full versions via their official sources: Kwai, Tiktok, and Movielens.

To facilitate this line of research, we provide some toy datasets:

• BaiduPan: Download (Extraction code: zsye)
• GoogleDrive: Download

If you need the full datasets, please contact the respective data owners.

Dataset Statistics

Dataset	#Interactions	#Users	#Items	Visual	Acoustic	Textual
Kwai	1,664,305	22,611	329,510	2,048	-	100
Tiktok	726,065	36,656	76,085	128	128	128
Movielens	1,239,508	55,485	5,986	2,048	128	100

File Descriptions

• train.npy: Train file. Each line is a user with her/his positive interactions with items (userID and micro-video ID).
• val.npy: Validation file. Each line is a user several positive interactions with items (userID and micro-video ID).
• test.npy: Test file. Each line is a user with several positive interactions with items (userID and micro-video ID).

---

Usage

The instruction of commands has been clearly stated in the codes. Run the following examples to train the models on different datasets.

Some important arguments:

• model_name: It specifies the type of model. Here we provide three options:

  1. MMGCN (by default) proposed in MMGCN: Multi-modal Graph Convolution Network for Personalized Recommendation of Micro-video, ACM MM2019. Usage: --model_name='MMGCN'
  2. VBPR proposed in VBPR: Visual Bayesian Personalized Ranking from Implicit Feedback, AAAI2016. Usage: --model_name 'VBPR'
  3. ACF proposed in Attentive Collaborative Filtering: Multimedia Recommendation with Item- and Component-Level Attention , SIGIR2017. Usage: --model_name 'ACF'
  4. GraphSAGE proposed in Inductive Representation Learning on Large Graphs, NIPS2017. Usage: --model_name 'GraphSAGE'
  5. NGCF proposed in Neural Graph Collaborative Filtering, SIGIR2019. Usage: --model_name 'NGCF'

• aggr_mode It specifics the type of aggregation layer. Here we provide three options:

  1. mean (by default) implements the mean aggregation in aggregation layer. Usage --aggr_mode 'mean'
  2. max implements the max aggregation in aggregation layer. Usage --aggr_mode 'max'
  3. add implements the sum aggregation in aggregation layer. Usage --aggr_mode 'add'

• concat: It indicates the type of combination layer. Here we provide two options:

  1. concat(by default) implements the concatenation combination in combination layer. Usage --concat 'True'
  2. ele implements the element-wise combination in combination layer. Usage --concat 'False'

-train.npy Train file. Each line is a user with her/his positive interactions with items: (userID and micro-video ID)
-val.npy Validation file. Each line is a user several positive interactions with items: (userID and micro-video ID)
-test.npy Test file. Each line is a user with several positive interactions with items: (userID and micro-video ID)

---

Acknowledgement

The copyright for the program is owned by Shandong University.

---

License

This program is licensed under the GNU General Public License 3.0. Any derivative work obtained under this license must be licensed under the GNU General Public License as published by the Free Software Foundation, either Version 3 of the License, or (at your option) any later version, if this derivative work is distributed to a third party.

For commercial projects that require the ability to distribute the code of this program as part of a program that cannot be distributed under the GNU General Public License, please contact weiyinwei@hotmail.com to purchase a commercial license.

Sources
1. https://github.com/Liuwq-bit/LightGT
2. https://github.com/Liuwq-bit/LightGT
3. https://github.com/weiyinwei/PHR_GCN

Citation

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

@inproceedings{MMGCN,  
  title     \= {MMGCN: Multi-modal graph convolution network for personalized recommendation of micro-video},  
  author    \= {Wei, Yinwei and Wang, Xiang and Nie, Liqiang and He, Xiangnan and Hong, Richang and Chua, Tat-Seng},  
  booktitle \= {Proceedings of the 27th ACM International Conference on Multimedia},  
  pages     \= {1437--1445},  
  year      \= {2019}  
}