An implementation trying to reproduce "Modeling Multi-Purpose Sessions for Next-Item Recommendations via Mixture-Channel Purpose Routing Networks" based on SR-GNN code.
The dataset is Yoochoose-buy prerpocessed by ourselves.
You can directly run the file 'main.py' to train and test the model.
For example: python main.py --dataset=yoochoose
Note that the result is not identical to the paper claimed performance.