You can get the data from amazon website and process it using the script
sh prepare_data.sh
Because getting and processing the data is time consuming，so we had processed it and upload it for you. You can unzip it to use directly.
tar -zxvf data.tar.gz
mv data/* .
When you see the files below, you can do the next work.
- cat_voc.pkl
- mid_voc.pkl
- uid_voc.pkl
- local_train_splitByUser
- local_test_splitByUser
python train.py train [model name]
The model blelow had been supported:
- DNN
- PNN
- Wide (Wide&Deep NN)
- DIN (https://arxiv.org/abs/1706.06978)
- DIEN (Our model)