![[2601.15673] Enhancing guidance for missing data in diffusion-based sequential recommendation](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2601.15673] Enhancing guidance for missing data in diffusion-based sequential recommendation
Summary
This paper presents the Counterfactual Attention Regulation Diffusion model (CARD) to improve sequential recommendation systems by addressing missing data issues and enhancing user guidance.
Why It Matters
As recommendation systems increasingly rely on user data, addressing the challenges posed by missing data is crucial for improving prediction accuracy. The CARD model offers a novel approach to enhance guidance, making it relevant for developers and researchers in AI and machine learning.
Key Takeaways
- CARD model improves sequential recommendations by focusing on critical turning points in user interest.
- Utilizes Dual-side Thompson Sampling to identify significant shifts in user preferences.
- Employs a counterfactual attention mechanism to enhance the quality of guidance signals.
- Demonstrates effectiveness on real-world data without high computational costs.
- Addresses a significant gap in existing recommendation methods by managing missing data.
Computer Science > Information Retrieval arXiv:2601.15673 (cs) [Submitted on 22 Jan 2026 (v1), last revised 13 Feb 2026 (this version, v2)] Title:Enhancing guidance for missing data in diffusion-based sequential recommendation Authors:Qilong Yan, Yifei Xing, Dugang Liu, Jingpu Duan, Jian Yin View a PDF of the paper titled Enhancing guidance for missing data in diffusion-based sequential recommendation, by Qilong Yan and 4 other authors View PDF HTML (experimental) Abstract:Contemporary sequential recommendation methods are becoming more complex, shifting from classification to a diffusion-guided generative paradigm. However, the quality of guidance in the form of user information is often compromised by missing data in the observed sequences, leading to suboptimal generation quality. Existing methods address this by removing locally similar items, but overlook ``critical turning points'' in user interest, which are crucial for accurately predicting subsequent user intent. To address this, we propose a novel Counterfactual Attention Regulation Diffusion model (CARD), which focuses on amplifying the signal from key interest-turning-point items while concurrently identifying and suppressing noise within the user sequence. CARD consists of (1) a Dual-side Thompson Sampling method to identify sequences undergoing significant interest shift, and (2) a counterfactual attention mechanism for these sequences to quantify the importance of each item. In this manner, CARD provides the...