DeepCoNN

This is a PyTorch implementation of DeepCoNN from the paper:

Lei Zheng, Vahid Noroozi, and Philip S Yu. 2017. Joint deep modeling of users and items using reviews for recommendation. In WSDM. ACM, 425-434.

Before Running Code

Get Data

Download and unzip "Digital Music" data set from :
http://jmcauley.ucsd.edu/data/amazon/
Then put it under the path data

Get Pre-trained Word Embedding Model

We use GoogleNews-vectors-negative300.bin as pre-trained word embedding model. You could find it at:
https://code.google.com/archive/p/word2vec/
Then put it under the path data

Environments

pandas~=1.0.3
numpy~=1.18.1
gensim~=3.8.0
pytorch~=1.3.1
nltk~=3.4.5
scikit-learn~=0.22.1

Train & Eval Model

Data Pre-processing

python -m utils.data_reader

Train Model

python train.py

You will find trained model file in model/checkpoints

Eval Model

Replace the model path in eval.py at first.

python eval.py