Multi-Perspective Attention Mechanism for Bias-Aware Sequential Recommendation

TL;DR

This work tackles bias-induced distortions in sequential recommendations by introducing MABSRec, a bias-aware framework that constructs three bias-specific views (popularity, amplified subjective bias, and debiased) and fuses them through an adaptive multi-bias attention mechanism. The model enriches item representations with graph convolutions per bias and encodes sequences via a shared Transformer module, enabling robust capture of dynamic user interests. Experimental results on Amazon Beauty, Amazon Sports, and MovieLens-20M show MABSRec consistently outperforms strong baselines in Recall@N and NDCG@N, with ablations confirming the importance of graph structure and adaptive bias fusion. The approach offers a practical pathway to mitigate the Matthew Effect in sequential recommendations and enhance bias-aware personalization at scale.

Abstract

In the era of advancing information technology, recommender systems have emerged as crucial tools for dealing with information overload. However, traditional recommender systems still have limitations in capturing the dynamic evolution of user behavior. To better understand and predict user behavior, especially taking into account the complexity of temporal evolution, sequential recommender systems have gradually become the focus of research. Currently, many sequential recommendation algorithms ignore the amplification effects of prevalent biases, which leads to recommendation results being susceptible to the Matthew Effect. Additionally, it will impose limitations on the recommender system's ability to deeply perceive and capture the dynamic shifts in user preferences, thereby diminishing the extent of its recommendation reach. To address this issue effectively, we propose a recommendation system based on sequential information and attention mechanism called Multi-Perspective Attention Bias Sequential Recommendation (MABSRec). Firstly, we reconstruct user sequences into three short types and utilize graph neural networks for item weighting. Subsequently, an adaptive multi-bias perspective attention module is proposed to enhance the accuracy of recommendations. Experimental results show that the MABSRec model exhibits significant advantages in all evaluation metrics, demonstrating its excellent performance in the sequence recommendation task.

Figures (10)

• Figure 1: Matthew Effect in Recommendation
• Figure 2: Framework diagram of the MABSRec model
• Figure 3: Calculation of the degree of bias
• Figure 4: Short sequence recombination
• Figure 5: graph structure message passing aggregation