/MPCN

MPCN implemented using Pytorch for KDD 2018 Paper "Multi-Pointer Co-Attention Networks for Recommendation"

Primary LanguagePython

MPCN

Implementation for the paper：
Tay, Yi, Anh Tuan Luu, and Siu Cheung Hui. "Multi-pointer co-attention networks for recommendation." In Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pp. 2309-2318. 2018.

Environments

  • python 3.8
  • pytorch 1.70

Dataset

You need to prepare the following documents:

  1. Dataset(data/Digital_Music.json.gz)
    Download from http://deepyeti.ucsd.edu/jianmo/amazon/index.html (Choose Digital Music)

  2. Word Embedding(embedding/glove.6B.50d.txt)
    Download from https://nlp.stanford.edu/projects/glove

Pre-Process

Preprocess origin dataset which is json format to be train.csv,valid.csv and test.csv.

python data_process.py

Running

Train and test:

python main.py --device cuda:0