FairSR: Fairness-aware Sequential Recommendation through Multi-Task Learning with Preference Graph Embeddings

Sequential recommendation (SR) learns from the temporal dynamics of user-item interactions to predict next ones. Fairness-aware mitigates a variety algorithmic biases in learning user preferences. This article aims at bringing marriage between SR and fairness. We propose novel fairness-aware sequential task, which new metric, interaction fairness , is defined estimate how recommended items are fairly interacted by users with different protected attribute groups. multi-task learning-based deep end-to-end model, FairSR, consists two parts. One learn distill personalized features given her item sequence for SR. The other preference graph embedding (FPGE). aim FPGE two-fold: incorporating knowledge users’ items’ attributes their correlation into entity representations, alleviating unfair distributions on items. Extensive experiments conducted three datasets show FairSR can outperform state-of-the-art models performance. In addition, also exhibit promising