search

RUCAIBox / NCL

[WWW'22] Official PyTorch implementation for "Improving Graph Collaborative Filtering with Neighborhood-enriched Contrastive Learning".
118 stars 20 forks source link

A small question about the reported number. #35

Closed SpaceLearner closed 1 year ago

SpaceLearner commented 1 year ago

Dear authors of NCL， I find two problems of the reported numbers of dataset Alibaba-iFashion in the paper:

image

Maybe you could make a refinement.

linzihan-backforward commented 1 year ago

Hi ! Thanks for your careful reading. They are all the typos in our paper. The right numbers are :

BPR Recall@10 0.0303 NeuMF Recall@10 0.0182 LightGCN NDCG@10 0.0246 LightGCN NDCG@20 0.0301

SpaceLearner commented 1 year ago

Thanks a lot for your in time reply.