DA2-Unlearn: Dual-Adaptive Forget-Repair-Based Recommendation Unlearning
Modern recommender systems increasingly face recommendation unlearning requests, where specific user-item interactions must be revoked due to interest drift, accidental feedback, data poisoning, or privacy concerns. Existing approaches overlook varying request difficulty and the trade-off between effective forgetting and model utility. We propose DA2-Unlearn, a dual-adaptive forget-repair framework. It uses intrinsic request difficulty to adapt forgetting strength, then applies gradient-guided forgetting-first optimization that emphasizes early removal before restoring recommendation utility. Experiments on real-world and simulated datasets demonstrate significantly better unlearning effectiveness than existing methods while maintaining comparable recommendation utility.