CUSTOM

This repo provides the code for reproducing the experiments in NLPCC2021 paper: CUSTOM: Aspect-Oriented Product Summarization for E-Commerce

Environment

torch==0.4.0

UniLM

• Train
  1. make pretrained_model folder, download Bert-base chinese version and move to this folder
  2. sh run_seq2seq.sh
• Predict
  1. sh run_decode.sh

Reference

Thanks for your citation:

@article{Liang2021CUSTOMAP,
  title={CUSTOM: Aspect-Oriented Product Summarization for E-Commerce},
  author={Jiahui Liang and Junwei Bao and Yifan Wang and Youzheng Wu and Xiaodong He and Bowen Zhou},
  journal={ArXiv},
  year={2021},
  volume={abs/2108.08010}
}

Contact

459605835@qq.com