CCIIPLab / MCCLK

The source code for "Multi-level Cross-view Contrastive Learning for Knowledge-aware Recommender System".

Multi-level Cross-view Contrastive Learning for Knowledge-aware Recommender System

This is our Pytorch implementation for the paper:

Ding Zou, Wei Wei, Xian-Ling Mao, Ziyang Wang, Minghui Qiu, Feida Zhu, Xin Cao (2022). Multi-level Cross-view Contrastive Learning for Knowledge-aware Recommender System, Paper in arXiv. In SIGIR'22.

Introduction

Multi-level Cross-view Contrastive Learning for Knowledge-aware Recommender System (MCCLK) is a knowledge-aware recommendation solution based on GNN and Contrastive Learning, proposing a multi-level cross-view contrastive framework to enhance representation learning from multi-faced aspects.

Requirement

The code has been tested running under Python 3.7.9. The required packages are as follows:

• pytorch == 1.5.0
• numpy == 1.15.4
• scipy == 1.1.0
• sklearn == 0.20.0
• torch_scatter == 2.0.5
• torch_sparse == 0.6.10
• networkx == 2.5

Usage

The hyper-parameter search range and optimal settings have been clearly stated in the codes (see the parser function in utils/parser.py).

• Train and Test

python main.py 

Citation

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

@inproceedings{mcclk2022,
  author    = {Zou, Ding and
               Mao, Xian-Ling and
	       Wang, Ziyang and
	       Qiu, Minghui and
	       Zhu, Feida and
	       Cao, Xin},
  title     = {Multi-level Cross-view Contrastive Learning for Knowledge-aware Recommender System},
  booktitle = {Proceedings of the 45th International {ACM} {SIGIR} Conference on
               Research and Development in Information Retrieval, {SIGIR} 2022, Madrid,
               Spain, July 11-15, 2022.},
  year      = {2022},
}

Dataset

We provide three processed datasets: Book-Crossing, MovieLens-1M, and Last.FM.

We follow the paper " Ripplenet: Propagating user preferences on the knowledge graph for recommender systems." to process data.

		Book-Crossing	MovieLens-1M	Last.FM
User-Item Interaction	#Users	17,860	6,036	1,872
	#Items	14,967	2,445	3,846
	#Interactions	139,746	753,772	42,346
Knowledge Graph	#Entities	77,903	182,011	9,366
	#Relations	25	12	60
	#Triplets	151,500	1,241,996	15,518

Reference

• We partially use the codes of KGIN.
• You could find all other baselines in Github.