Skip to content

Latest commit

 

History

History
25 lines (20 loc) · 818 Bytes

README.md

File metadata and controls

25 lines (20 loc) · 818 Bytes
Raw

GraphRec

This is our implementation for the recsys 2019 paper:

Rashed, Ahmed, Josif Grabocka, and Lars Schmidt-Thieme. "Attribute-aware non-linear co-embeddings of graph features."13th ACM Conference on Recommender Systems (RecSys). 2019.

Enviroment

* pandas==1.0.3
* tensorflow==1.14.0
* matplotlib==3.1.3
* numpy==1.18.1
* six==1.14.0
* scikit_learn==0.23.1

Steps

  1. Uncomment the respective code of the dataset you want to reproduce the results for and run "python GraphRec.py".

Paper

Preprint version :https://www.ismll.uni-hildesheim.de/pub/pdfs/Ahmed_RecSys19.pdf

Supplementary/Extra Results

ML100k Experiment using the u1.base/u1.test splits

Model RMSE
GraphRec (w/ Graph Feat.) 0.904
GraphRec (w/ Graph Feat. & Users/Items Attributes) 0.897