The official code of Recsys'25 paper 'USB-Rec: An Effective Framework for Improving Conversational Recommendation Capability of Large Language Model'
https://arxiv.org/pdf/2509.20381
Set the model in SESConfig you want to use in ses_functions.py, which can be a locally deployed LLM or an open-sourced LLM.
Then set the api_key and base_url in create_client function if you use an open-sourced LLM.
run podcs_main.ipynb and results will be saved in po_results.jsonl.
We deploy llama-factory to training the LLMs.
run ses_main.ipynb and results will be saved in results.jsonl.
We use the dataset from iEval.
Please cite the following paper if you find our code helpful.
@inproceedings{wen2025usb,
title={USB-Rec: An Effective Framework for Improving Conversational Recommendation Capability of Large Language Model},
author={Wen, Jianyu and Wang, Jingyun and Yan, Cilin and Cai, Jiayin and Jiang, Xiaolong and Zhang, Ying},
booktitle={Proceedings of the Nineteenth ACM Conference on Recommender Systems},
pages={472--481},
year={2025}
}