README.md

Coupled Variational Recurrent Collaborative Filtering

Code of KDD 2019 paper Coupled Variational Recurrent Collaborative Filtering.

First Step:

Download Data from https://drive.google.com/drive/folders/1HeudBdACXQHSYTb04fNQNk7mVxiBGq7C?usp=sharing and put them in the 'data' folder. Each txt file has six columns as follows:

36955 21  3 789652009 0 0
36955 47  5 789652009 789652009 0
36955 1058  3 789652009 789652009 0
35139 1 4 822873600 0 0
35139 10  4 822873600 822873600 0
35435 10  3 822873600 0 822873600
35435 11  4 822873600 822873600 0
35139 18  4 822873600 822873600 0
35139 19  4 822873600 822873600 0
35621 19  1 822873600 0 822873600
    ...

The six columns correspond to:

user_id  item_id  rating  timestamp user_last_rating_timestamp  item_last_rated_timestamp

Each file is sorted based on the timestamp in the fourth column. The '0' elements in the fifth or the sixth column represent the new users or new items.

Second Step:

In command line run:

sh ml_10M.sh

Cite this work

Qingquan Song, Shiyu Chang, and Xia Hu. "Coupled Variational Recurrent Collaborative Filtering." Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. ACM, 2019. (Download)

Biblatex entry:

@inproceedings{song2019coupled,
  title={Coupled Variational Recurrent Collaborative Filtering},
  author=Song, Qingquan and Chang, Shiyu and Hu, Xia},
  booktitle={Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining},
  pages={335--343},
  year={2019},
  organization={ACM}
}