Self-Supervised pre-training model based on Multi-view for MOOC Recommendation

Runyu Tian, Juanjuan Cai, Chuanzhen Li, Jingling Wang

Abstract: Highlights•Prerequisite dependency of attributes elicits semantic associations of course data.•Correlations between multi-view are treated as self-supervised signals.•Pretrain the model to enhance representations through contrastive learning with MIM.•Finetune pretrained encoder for downstream task to achieve final recommendation.