Latent Relational Metric Learning via Memory-based Attention for Collaborative Ranking

Implements a barebones reference implementation of the LRML model used in our WWW'18 experiments.

Note: Due to high volume of requests, I am releasing this source code, (heavily cleaned to remove other irrelevant WIP models) about 1+ years after I done the experiments. Please check the detailed comments in the model file.

Reference

If you find our model useful, please consider citing our work.

@inproceedings{DBLP:conf/www/TayTH18,
  author    = {Yi Tay and
               Luu Anh Tuan and
               Siu Cheung Hui},
  title     = {Latent Relational Metric Learning via Memory-based Attention for Collaborative
               Ranking},
  booktitle = {Proceedings of the 2018 World Wide Web Conference on World Wide Web,
               {WWW} 2018, Lyon, France, April 23-27, 2018},
  pages     = {729--739},
  year      = {2018},
  crossref  = {DBLP:conf/www/2018},
  url       = {http://doi.acm.org/10.1145/3178876.3186154},
  doi       = {10.1145/3178876.3186154},
  timestamp = {Mon, 16 Apr 2018 08:43:10 +0200},
  biburl    = {https://dblp.org/rec/bib/conf/www/TayTH18},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}