Bundle MCR Framework

This is the pytorch implementation of this paper Bundle MCR: Towards Conversational Bundle Recommendation.

Zhankui He, Handong Zhao, Tong Yu, Sungchul Kim, Fan Du, Julian McAuley. 16th ACM Conference on Recommender Systems (RecSys '22). Oral.

Bunt Implementation

NOTE: The details of all dataset processing (and more information) are in Appendix.pdf.

We use python 3.6 and other python dependencies are listed in requirements.txt, you can install them with pip install -r requirements.txt.

Quick Start

1. Offline Pre-Training: Use bash scripts/train_offline.sh ${device_id} ${seed}, where ${device_id} is used to specify your GPU id, and ${seed} is the random seed you assign. For example:

   bash scripts/train_offline.sh 0 0
   

2. Online Fine-Tuning: Use bash scripts/train_online.sh ${device_id} ${seed} ${pre_trained_model_path}. The explanation of arguments are as the same as step 1, except for ${pre_trained_model_path}, which is the *.pt model path to load as pre-trained Bunt for online fine-tuning, which can be found in checkpoints folder by default. For example:

   bash scripts/train_online.sh 0 0 checkpoints/steam/model_1.pt
   

3. Collect Results: You are free to print out your results using python tools/results.py ${ckpt_path}, where ${ckpt_path} is the path of your experiment folder, such as checkpoints/steam.

Data Processing (Use Steam dataset as an example)

1. Go to steam folder, cd raw/;
2. For data interaction processing and interactions splitting, use python 0_data_splitting.py
3. To process attributes for Bundle MCR, use python 1_item_attr.py;
4. To precess categories for Bundle MCR, use python 2_item_cate.py.

Data Downloading (Updated on 12/09/22)

Someone encountered the issues of downloading datasets from Git LFS, therefore we also upload the experimental datasets (raw, processed and related pre-processing scripts) to Google Drive. Please check this link to download those datasets.

Bibtex

Please cite our paper if using this code, and feel free to contact [email protected] if any questions.

@inproceedings{he22bundle,
  title = "Bundle MCR: Towards conversational bundle recommendation",
  author = "Zhankui He and Handong Zhao and Tong Yu and Sungchul Kim and Fan Du and Julian McAuley",
  year = "2022",
  booktitle = "RecSys"
}