Decoupled Side Information Fusion for Sequential Recommendation

Abstract

Side information fusion for sequential recommendation (SR) aims to effectively leverage various side information to enhance the performance of next-item prediction. Most state-of-the-art methods build on self-attention networks and focus on exploring various solutions to integrate the item embedding and side information embeddings before the attention layer. However, our analysis shows that the early integration of various types of embeddings limits the expressiveness of attention matrices due to a rank bottleneck and constrains the flexibility of gradients. Also, it involves mixed correlations among the different heterogeneous information resources, which brings extra disturbance to attention calculation. Motivated by this, we propose Decoupled Side Information Fusion for Sequential Recommendation (DIF-SR), which moves the side information from the input to the attention layer and decouples the attention calculation of various side information and item representation. We theoretically and empirically show that the proposed solution allows higher-rank attention matrices and flexible gradients to enhance the modeling capacity of side information fusion. Also, auxiliary attribute predictors are proposed to further activate the beneficial interaction between side information and item representation learning. Extensive experiments on four real-world datasets demonstrate that our proposed solution stably outperforms state-of-the-art SR models. Further studies show that our proposed solution can be readily incorporated into current attention-based SR models and significantly boost performance.

Towards trustworthy recommenders: building explainable and unbiased recommendation systems

“…However, when using long-tail data to train conventional models, the common training approaches make the model prefer to recommend head items, which leads to popularity bias and hurts user experience due to recommending repeat items [65]. Many solutions have been proposed to mitigate popularity bias or long-tail problems in recommendation [15,63,276].…”

“…Besides, some other strategies leverage side information [276] or a memory bank [62] to improve the quality of tail items. Park and Tuzhilin [16] proposed a Clustered Tail (CT) method to group tail items using clustering methods and then recommended tail items based on the ratings in the clusters.…”

“…This kind of attribute information has been demonstrated as conveying useful information and contributing to user representation learning [386]. Besides, these studies have also illustrated that incorporating attribute information can mitigate the long-tail problem [15,276]…”