Recommending cold-start items is a long-standing and fundamental challenge inrecommender systems . CFscheme fails to use collaborative signals to infer user preference on these items . CLCRec aims to maximize themutual dependencies between item content and collaborative signals . It allows us to preserve collaborative signals in the content representations in thecontent representations for both warm and cold start items . It is a simpleyet effective Contrastive Learning-based Cold-start Recommendationframework(CLCRec) It is based on contrastive learning and has significant improvements over state-of-the-art approaches in both warm-and-cold-start scenarios in both scenarios . It has three components:contrastive pair organization, contrastive embedding, and contrastiveoptimization modules. It is an effective tool that preserves collaborative signals for bothwarm and cold-starting items. It allows for us to preserved collaborative signals. It has been tested on four publicly accessible datasets. Through extensiveexperiments on four datasets, we observe that CLCReachieves significant improvements. We observe that it has significantly improved .

Author(s) : Yinwei Wei, Xiang Wang, Qi Li, Liqiang Nie, Yan Li, Xuanping Li, Tat-Seng Chua

Links : PDF - Abstract

Code :
Coursera

Keywords : cold - signals - start - contrastive - collaborative -

Leave a Reply

Your email address will not be published. Required fields are marked *