Integrating Item Relevance in Training Loss for Sequential Recommender Systems
Andrea Bacciu, Federico Siciliano, Nicola Tonellotto, Fabrizio Silvestri
TL;DR
This paper tackles noise in Sequential Recommender Systems by introducing a Multi Future Items evaluation protocol and a Relevance-based loss that trains with multiple future items. By weighting future interactions according to a decreasing relevance function, the authors show robust improvements over a single-item evaluation, with notable gains in $NDCG$@10 and $HR$@10 across multiple datasets using the SASRec baseline. The key contributions are the MFI evaluation framework, the formalization of item relevance functions (Fixed, Linear, Power, Exponential), and the Relevance-based loss that incorporates these weights into training. The approach enhances robustness to noisy sequences and offers practical implications for improving ranking performance in real-world SRS deployments.
Abstract
Sequential Recommender Systems (SRSs) are a popular type of recommender system that learns from a user's history to predict the next item they are likely to interact with. However, user interactions can be affected by noise stemming from account sharing, inconsistent preferences, or accidental clicks. To address this issue, we (i) propose a new evaluation protocol that takes multiple future items into account and (ii) introduce a novel relevance-aware loss function to train a SRS with multiple future items to make it more robust to noise. Our relevance-aware models obtain an improvement of ~1.2% of NDCG@10 and 0.88% in the traditional evaluation protocol, while in the new evaluation protocol, the improvement is ~1.63% of NDCG@10 and ~1.5% of HR w.r.t the best performing models.