Popularity-aware sequential recommendation with user desire

Jiajin Wu, Bo Yang, Runze Mao, Qing Li

Abstract: Highlights•A popularity-aware sequential recommendation model with user desire is proposed.•A new causal graph is proposed to model both the popularity and the user desire.•Theoretical intervention analysis is conducted to get debiased recommendation.•The proposed model can both alleviate and leverage popularity bias as needed.•Experiments demonstrate the proposed model achieves state-of-the-art performance.