Sep 2026· ACM Transactions on Information Systems· 0 citations· 126 references
TL;DR
This work introduces ScoreRec, a novel sequential recommendation framework that reformulates the recommendation task as a continuous-time score-based generative process that leverages a stochastic differential equation with a continuous-time noise schedule to mitigate the mode-seeking bias.
Abstract
Sequential recommendation generates personalized item rankings by estimating latent user-preference distributions from historical interactions. Although Diffusion Models (DMs) have gained prominence due to their exceptional capacity to capture complex distributions, current DM-based recommenders exhibit a critical limitation Distributional Misalignment, where mode-seeking denoising processes disproportionately allocate probability mass to popular items, thereby diverging from users’ genuine preferences. To bridge this gap, we introduce ScoreRec, a novel sequential recommendation framework that reformulates the recommendation task as a continuous-time score-based generative process. Unlike conventional discrete denoising diffusion probabilistic models (DDPMs) that rely on fixed time-step approximations, ScoreRec models the diffusion process via a Stochastic Differential Equation. This formulation allows ScoreRec to estimate the gradient of the log-density with infinite temporal resolution, effectively reallocating probability mass to low-frequency yet personally relevant items and thereby mitigating the mode-seeking bias. Specifically, ScoreRec leverages a stochastic differential equation with a continuous-time noise schedule, ensuring that the noise level is dynamically adjusted to the local embedding density. During inference, an adaptive-step-size solver integrates the corresponding probability-flow ordinary differential equation to deterministically generate recommendations, striking a principled balance between accuracy and efficiency. Comprehensive experiments on three public benchmarks demonstrate that ScoreRec consistently surpasses eleven state-of-the-art baselines in both accuracy and diversity. Our code is publicly available at: https://github.com/AONE-NLP/ScoreRec.
A user history can admit multiple plausible next interactions, yet conventional sequential recommenders produce a single deterministic ranking. While diffusion models offer a natural way to sample alternative continuations, we show that existing target-item diffusion recommenders often suffer from severe sampling colla...
Louis Castres Saint Martin, Alain Rakotomamonjy, P. Gallinari· 0 citations
Sequential recommendation must infer evolving preferences from interaction histories that are sparse, temporally irregular, and heterogeneous in relevance. Existing diffusion-based recommenders improve representation generation and recovery, yet their perturbation strength is typically governed by the diffusion step an...
FairDiff is proposed, a plug-and-play fairness-aware diffusion framework that achieves state-of-the-art performance while effectively mitigating the self-reinforcing Matthew Effect, highlighting its value as a general framework for DRMs.
Song-Li Wu, Xian-Quan Wang, Zhao-Cheng Du et al.· 2 citations
A debiasing framework, termed Mult-IPS, is introduced, which integrates multinomial likelihood with IPS to capture global and unbiased user preferences over the entire item set, and a Bi-Weighting strategy that jointly leverages propensity scores and a collection model, incorporating a smoothing mechanism to enhance th...
Tian-Yu Zhu, Jian-Dong Ding, Yan-Song Shi et al.· ACM Transactions on Informat...· 0 citations
Popularity bias is a well-known challenge in recommender systems, often leading to the over-exposure of popular items and reduced catalog coverage. In this work, we propose PR-entmax, a sparse prior-regularized objective that makes popularity correction context-dependent. We derive our method from a generalized variati...
Anton Pembek, Anton Klenitskiy, Alexander Savchenko et al.· Proceedings of the 35th ACM...· 0 citations
DynFreqRec, a Frequency-Aware Training–Inference Synergistic Enhancement Framework for sequential recommendation, builds on an existing Base Time-Frequency Encoder that uses adaptive frequency-domain filtering and the Haar Wavelet Transform to characterize global preferences, periodic interests, and local non-stationar...