CR-Walker

Code implementation for our paper "Bridging the Gap between Conversational Reasoning and Interactive Recommendation".

you can find our paper at arxiv.

Cite this paper:

@article{ma2020bridging,
      title={Bridging the Gap between Conversational Reasoning and Interactive Recommendation}, 
      author={Wenchang Ma and Ryuichi Takanobu and Minghao Tu and Minlie Huang},
      journal={arXiv preprint arXiv:2010.10333},
      year={2020}
}

Data

google link to raw data and our model checkpoints. Zipped content:

CR-Walker
├─data
│  ├─gorecdial
│  │  └─raw
│  ├─gorecdial_gpt
│  ├─redial
│  │  └─raw
│  └─redial_gpt
└─saved

download and unzip to [your home directory]/CR-Walker/.

Train

For GoRecdial:

python train_gorecdial.py --option train --model_name <your_model_name>

For Redial:
```
python train_redial.py --option train --model_name <your_model_name> --pretrain
```
*We implemented an MIM pretraining stage similar to KGSF to accelerate training.

Test

For GoRecdial

python train_gorecdial.py --option test --model_name gorecdial_128

For Redial:

python train_redial.py --option test --model_name redial_128

You can directly evaluate the best model checkpoints for the two datasets that we provided. The results may slightly differ from the paper since we re-trained the model.

Requirements

python==3.6.10

pytorch==1.4.0

torch_geometric==1.6.0# example1

dumbdumbdumb/example1

CR-Walker

Data

Train

Test

Requirements