Collaborative Memory Network for Recommendation Systems

Implementation for

Travis Ebesu, Bin Shen, Yi Fang. Collaborative Memory Network for Recommendation Systems. In Proceedings of the 41st International ACM SIGIR Conference on Research and Development in Information Retrieval, 2018.

https://arxiv.org/pdf/1804.10862.pdf

Bibtex

@inproceedings{Ebesu:2018:CMN:3209978.3209991,
 author = {Ebesu, Travis and Shen, Bin and Fang, Yi},
 title = {Collaborative Memory Network for Recommendation Systems},
 booktitle = {The 41st International ACM SIGIR Conference on Research \&\#38; Development in Information Retrieval},
 series = {SIGIR '18},
 year = {2018},
 isbn = {978-1-4503-5657-2},
 location = {Ann Arbor, MI, USA},
 pages = {515--524},
 numpages = {10},
 url = {http://doi.acm.org/10.1145/3209978.3209991},
 doi = {10.1145/3209978.3209991},
 acmid = {3209991},
 publisher = {ACM},
 address = {New York, NY, USA},
 keywords = {collaborative filtering, deep learning, memory networks},
} 

Running Collaborative Memory Network

python train.py --gpu 0 --dataset data/citeulike-a.npz --pretrain pretrain/citeulike-a_e50.npz

To pretrain the model for initialization

python pretrain.py --gpu 0 --dataset data/citeulike-a.npz --output pretrain/citeulike-a_e50.npz

Requirements

• Python 3.6
• tensorflow==1.14
• dm-sonnet==1.25
• tensorflow_probability==0.7
• tqdm

Data Format

The structure of the data in the npz file is as follows:

train_data = [[user id, item id], ...]
test_data = {userid: (pos_id, [neg_id1, neg_id2, ...]), ...}