DIREC

Code

This is the source code for the Paper: Enhancing Session-based Recommendation with Contrastive Learning from Intent and Attribute Views by Exploring Substitutable and Complementary Relations.

Requirements

Python 3.9
PyTorch 1.12

Best Hyperparameter:

Tmall: alpha=0.02, beta=100, num_coarse_sampling=90
E-Commerce (CIKM19): alpha=0.02, beta=150, num_coarse_sampling=90
Cosmetics: alpha=0.02, beta=100, num_coarse_sampling=90

Datasets:

Dataset	URLs
Tmall	https://tianchi.aliyun.com/dataset/dataDetail?dataId=42
Cosmetics	https://www.kaggle.com/datasets/mkechinov/ecommerce-events-history-in-cosmetics-shop
E-Commerce	https://tianchi.aliyun.com/competition/entrance/231721/information

Train & Test:

For Tmall and E-Commerce(CIKM19):

python main.py --dataset Tmall --alpha 0.02 --beta 100
python main.py --dataset CIKM19 --alpha 0.02 --beta 150

For Cosmetics:

python main_cosmetics.py --dataset Cosmetics --alpha 0.02 --beta 100

Preprocessed data:

The datasets have been preprocessed and encoded with pickle, which can be downloaded from the link (password: v3bj)