An Unsupervised Aspect-aware Recommendation Model withExplanation Text Generation

This repository is the official implementation of [An Unsupervised Aspect-aware Recommendation Model withExplanation Text Generation]. Besides, we also provide the implementation of other baseline models which are used in our paper.

Requirements

To install requirements:

pip install -r requirements.txt

Datasets

The dataset is not contained in this repository. You can download the datasets here:

• Amazon Pet Supplies Password: qf3s

Training Procedures

Preparation:

1. git clone this repository
2. download the datasets
3. place the datasets to the data directory

Comparative baseline models:

1. cd src/amazon_pet_supplies/model_name
2. python train_mdel.py

U-ARM:

• Pretraining the ABAE model:

  1. cd src/amazon_pet_supplies/abae
  2. python train_abae.py

• Pretraining the NCF model:

  1. cd src/amazon_pet_supplies/ncf
  2. python train_ncf.py

• Pretraining the U-ARM rating prediction module:

  1. cd src/amazon_pet_supplies/u_arm_rating
  2. python train_uarm.py

• Training the U-ARM explanation text generation module:

  1. cd src/amazon_pet_supplies/u_arm_review
  2. python train_uarm.py