<aside> 💡 Transferability : 다양한 Domain과 system으로 transfer learning이 가능한 CLUE

</aside>

KEYWORDS

1. universal user representation
2. recommender system
3. contrastive learning
4. transfer learning

1 INTRODUCTION

Studies on the scaling and generalization ability of general-purpose pretrained user representation learning still remain underexplored for various downstream recommendation tasks

• 1. Can general-purpose user representations learned from multiple source data provide a promising transfer learning capability?
• 2. Are pretraining and downstream task performances positively correlated?

  

• 3. How various tasks can the pretrained user representations address?
• 4. Does scaling up the pretraining model improve the generalization performance?