S 2 CDR: Smoothing-Sharpening Process Model for Cross-Domain Recommendation Conference: Proceedings of the ACM Web Conference 2026; April 13–17, 2026; Dubai, United Arab Emirates Proceedings of the ACM Web Conference 2026 (WWW ’26), April 13–17, 2026, Dubai, United Arab Emirates DOI: 10.1145/3774904.3792189 ISBN: 979-8-4007-2307-0/2026/04 CCS: Information systems Recommender systems

Xiaodong Li, Juwei Yue, Xinghua Zhang, Jiawei Sheng, Wenyuan Zhang, Taoyu Su, Zefeng Zhang, Tingwen Liu · 2026

Read S 2 CDR: Smoothing-Sharpening Process Model for Cross-Domain Recommendation Conference: Proceedings of the ACM Web Conference 2026; April 13–17, 2026; Dubai, United Arab Emirates Proceedings of the ACM Web Conference 2026 (WWW ’26), April 13–17, 2026, Dubai, United Arab Emirates DOI: 10.1145/3774904.3792189 ISBN: 979-8-4007-2307-0/2026/04 CCS: Information systems Recommender systems by Xiaodong Li, Juwei Yue, Xinghua Zhang, Jiawei Sheng, Wenyuan Zhang, Taoyu Su, Zefeng Zhang, Tingwen Liu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-02725.txt.

Introduction

Xiaodong Li, Juwei Yue, Xinghua Zhang, and colleagues help new users of one service by borrowing taste signals from another. Their 2026 model, S2CDR, addresses cross-domain recommendation for cold-start users with no local history. Diffusion-based rivals transfer preferences well but ignore links between items across domains, and their added noise can blur personal taste.

The new design replaces noise with a two-way flow. A smoothing step spreads interaction data across an item-similarity graph using a heat-style rule and a filter that keeps the clean signal. A sharpening step then rebuilds the missing interactions for new users, guided by differential equations. Across three real test cases, the training-free method beat the best prior systems.

Catalog record

CREATORXiaodong Li, Juwei Yue, Xinghua Zhang, Jiawei Sheng, Wenyuan Zhang, Taoyu Su, Zefeng Zhang, Tingwen Liu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE55720
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.02725

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

S 2 CDR: Smoothing-Sharpening Process Model for Cross-Domain Recommendation Conference: Proceedings of the ACM Web Conference 2026; April 13–17, 2026; Dubai, United Arab Emirates Proceedings of the ACM Web Conference 2026 (WWW ’26), April 13–17, 2026, Dubai, United Arab Emirates DOI: 10.1145/3774904.3792189 ISBN: 979-8-4007-2307-0/2026/04 CCS: Information systems Recommender systems Xiaodong Li, Juwei Yue, Xinghua Zhang, Jiawei Sheng, Wenyuan Zhang, Taoyu Su, Zefeng Zhang, Tingwen Liu Abstract . User cold-start problem is a long-standing challenge in recommendation systems. Fortunately, cross-domain recommendation (CDR) has emerged as a highly effective remedy for the user cold-start challenge, with recently developed diffusion models (DMs) demonstrating exceptional performance. However, these DMs-based CDR methods focus on dealing with user-item interactions, overlooking correlations between items across the source and target domains. Meanwhile, the Gaussian noise added in the forward process of diffusion models would hurt user’s personalized preference, leading to the difficulty in transferring user preference across domains. To this end, we propose a novel paradigm of S moothing- S harpening Process Model for CDR to cold-start users, termed as S 2 CDR which features a corruption-recovery architecture and is solved with respect to ordinary differential equations (ODEs). Specifically, the smoothing process gradually corrupts the original user-item/item-item interaction matrices derived from both domains into smoothed preference signals in a noise-free manner, and the sharpening process iteratively sharpens the preference signals to recover the unknown interactions for cold-start users. Wherein, for the smoothing process, we introduce the heat equation on the item-item similarity graph to better capture the correlations between items across domains, and further build the tailor-designed low-pass filter to filter out the high-frequency noise information for capturing user’s intrinsic preference, in accordance with the graph signal processing (GSP) theory. Extensive experiments on three real-world CDR scenarios confirm that our S 2 CDR significantly outperforms previous SOTA methods in a training-free manner. † † cc-license: by 1. Introduction Recommendation systems (RS) ( 26 ; 35 ; 9 ) have been extensively applied in various online platforms, such as Alibaba (e-commerce) and TikTok (online video), leading to multiple user interactions in different categories on e-commerce websites and different media forms in online video. Consequently, developing effective strategies for providing recommendations to newly arrived (cold-start) users in various domains has emerged as a critical challenge in RS. To alleviate the cold-start problem ( 39 ; 32 ) , cross-domain recommendation (CDR) ( 17 ; 19 ; 23 ; 20 ) has received considerable attention in recent years, which aims to explore and exploit meaningful knowledge in related domains for achieving promising recommendation performance for cold-start users in another domain. A common idea of CDR methods is to utilize the overlapping users to bridge the source and target domains, and thus make recommendation for cold-start (non-overlapping) users who only have interactions in one domain. Following the above idea, traditional CDR methods ( 25 ; 14 ; 23 ) mostly adhere to the mapping-based paradigm, which encodes user/item representations in both domains separately, and then learns a mapping function to transfer user preference across domains. In addition, some meta-based methods ( 45 ; 17 ; 31 ) further adopt the meta-learning paradigm, treating each user’s CDR as an individual task and learning mapping functions for each user to achieve personalized preference transfer. Diffusion models (DMs) ( 10 ) have achieved remarkable success in the field of RS ( 34 ; 41 ; 40 ; 13 ; 24 ) due to their representation reconstruction ability. More recently, DMCDR ( 19 ) pioneers the application of diffusion models to CDR, which gradually corrupts the user representation by adding Gaussian noise in the forward process, and then iteratively generates personalized user representation in the target domain under the guidance of user preference derived from user-item interactions in the source domain. However, the DMs-based paradigm has the following two limitations for CDR task: (1) Lack of Item Correlation : the diffusion model emphasizes the user-item interactions, neglecting to capture the correlation between items across the source and target domains, which also serves as a crucial bridge for preference transfer. (2) Detriment of External Noise : the Gaussian noise added in the forward process would hurt user’s intrinsic personalized preference, leading to transfer the sub-optimal user preference across domains. To address the above issues, this paper proposes S 2 CDR , a novel S moothing- S harpening Process Model for CDR to cold-start users, which possesses a corruption-recovery architecture. Specifically, the smoothing process gradually smooths the user/item information including both user-item and item-item interaction signals from both domains in a noise-free manner with tailor-designed graph filters, supported by graph signal processing (GSP) ( 29 ; 21 ; 36 ; 38 ) . Wherein, we calculate the heat equation on the item-item similarity graph to capture the correlations between items across domains, and further build the effective low-pass filter to filter out the high-frequency information which is typically noise ( 21 ; 29 ) . Afterwards, the sharpening process iteratively refines and sharpens preference signals to describe user’s personalized preference, thus recovering the underlying interactions for cold-start users. To run our smoothing-sharpening process model, we further introduce ordinary differential equations (ODEs) ( 5 ; 3 ) for more effective and efficient cross-domain preference transfer via solving our smoothing and sharpening functions in a continuous manner, naturally characterizing the user’s intrinsic preference for transfer and contributing to our training-free method. Extensive experiments demonstrate that our S 2 CDR outperforms previous SOTA methods in three real-world CDR scenarios. Meanwhile, as a non-parametric method, our S 2 CDR shows high computational efficiency. Our main contributions can be summarized as follows: • We inspire a corruption-recovery paradigm for CDR, which fulfills the bridging effect of item correlations in cross-domain preference transfer and retains user’s intrinsic preference features. • We propose a novel Smoothing-Sharpening Process Model S 2 CDR for CDR to cold-start users, where the smoothing process smooths the user-item/item-item interaction signals for capturing the user preference based on tailor-designed graph filters, and the sharpening process sharpens the preference signals to recover unknown interactions for cold-start users. • We conduct extensive experiments in three real-world CDR scenarios to demonstrate the effectiveness and efficiency of our S 2 CDR over previous SOTA methods. Despite the fact that S 2 CDR does not require any training, we still achieve an average relative improvement of 7.07%, 7.42% with respect to HR and NDCG metrics over six cross-domain pairs. 2. Preliminary In this section, we review the background knowledge of graph signal processing (GSP), ordinary differential equations (ODEs), and defines our CDR problem settings. We also review diffusion models (DMs) and give a formal comparison between DMCDR and our S 2 CDR in Appendix A.1 . 2.1. Graph Signal Processing In this subsection, we introduce basic concepts of graph signal processing (GSP) ( 6 ; 27 ) , including graph signal, graph filter, and graph convolution. 2.1.1. Graph signal Given a weighted undirected graph \mathcal{G}=(\mathcal{V},\mathcal{E}) , where \mathcal{V}=\{v_{1},v_{2},\dots,v_{n}\} represents the vertex set with n nodes ( e.g. , users or items), and \mathcal{E} is the set of edges ( e.g. , user-item interactions). The graph \mathcal{G} induces an adjacency matrix \bm{A}\in\mathbb{R}^{n\times n} , and if there is an edge between node v_{i} and node v_{j} ,

Read the complete text