This repository provides the official implementation of HSMJ, as described in the ECML-PKDD 2026 paper "Hierarchically Structured Multi-Interest Joint Learning for Sequential Recommendation".

• HSMJ.py: implementation of the proposed HSMJ model;
• HSMJGCNNORQ.py: implementation of the ablated variant corresponding to Sub(4) in the experiments;
• MGNM.py: implementation of the MGNM baseline used for efficiency and performance comparison.

We conduct experiments on three subsets of Amazon review data, including Musical Instruments, Toys and Games, and Beauty.

A summary of dataset statistics is provided in the paper.

Dependencies are listed in requirements.txt.

pip install -r requirements.txt

Example training command:

python src/main.py --model_name <model_name> --dataset <dataset_name> --args ...

Please see Hyperparameter Setup and Ablation Study sections in the paper for details.