Zhongxuan Han
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Zhongxuan Han has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
- Chaochao Chen
Zhejiang University
7 shared publications - Xiaolin Zheng
Zhejiang University
6 shared publications - Yuyuan Li
Hangzhou Dianzi University, Hangzhou, China
5 shared publications - Jun Wang
OPPO Research Institute, Shenzhen, Guangdong, China
2 shared publications - Li Zhang
Zhejiang University, Hangzhou, China
2 shared publications - Weiming Liu
Zhejiang University
2 shared publications
Latest Publications
- Open Access
- 12 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 6021-6032
PRISM: Personalized Recommendation via Information Synergy Module
- 12 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 6021-6032
Open AccessPRISM: Personalized Recommendation via Information Synergy Module
- 8 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Services Computing
- Vol. 18 (5) , 3307-3320
Towards Fairness Exploration and Optimization for Digital Service Networks
Digital service networks often face the challenge of Service-Oriented Fairness (SOF), where service nodes with varying levels of activity may receive unequal treatment. This paper takes the recommendation service system as a representative case to explore and mitigate the impact of SOF. The SOF issue in the recommendation service system can be abstracted as User-Oriented Fairness (UOF), where service models often exhibit bias toward a small group of users, resulting in significant unfairness in the quality of recommendations. Existing research on UOF faces three major limitations, and no single approach effectively addresses all of them. Limitation 1: Post-processing methods fail to address the root cause of the UOF issue. Limitation 2: Some in-processing methods rely heavily on unstable user similarity calculations under severe data sparsity problems. Limitation 3: Other inprocessing methods overlook the disparate treatment of individual users within user groups. In this paper, we propose a novel Individual Reweighting for User-Oriented Fairness framework, namely IR-UOF, to address all the aforementioned limitations. The motivation behind IR-UOF is to introduce an in-processing strategy that addresses the UOF issue at the individual level without the need to explore user similarities. We first conduct extensive experiments on three real-world recommendation service datasets using four backbone recommendation models to demonstrate the effectiveness of IR-UOF in mitigating UOF and improving recommendation fairness. Furthermore, we select two general digital service datasets to prove that IR-UOF can be extended to tackle the general SOF issue in other types of digital service networks. In summary, the IR-UOF framework achieves optimal model performance across all datasets, while improving fairness by at least 3.8% in recommendation systems and 24.7% in general service systems.
- 8 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Services Computing
- Vol. 18 (5) , 3307-3320
Towards Fairness Exploration and Optimization for Digital Service Networks
Digital service networks often face the challenge of Service-Oriented Fairness (SOF), where service nodes with varying levels of activity may receive unequal treatment. This paper takes the recommendation service system as a representative case to explore and mitigate the impact of SOF. The SOF issue in the recommendation service system can be abstracted as User-Oriented Fairness (UOF), where service models often exhibit bias toward a small group of users, resulting in significant unfairness in the quality of recommendations. Existing research on UOF faces three major limitations, and no single approach effectively addresses all of them. Limitation 1: Post-processing methods fail to address the root cause of the UOF issue. Limitation 2: Some in-processing methods rely heavily on unstable user similarity calculations under severe data sparsity problems. Limitation 3: Other inprocessing methods overlook the disparate treatment of individual users within user groups. In this paper, we propose a novel Individual Reweighting for User-Oriented Fairness framework, namely IR-UOF, to address all the aforementioned limitations. The motivation behind IR-UOF is to introduce an in-processing strategy that addresses the UOF issue at the individual level without the need to explore user similarities. We first conduct extensive experiments on three real-world recommendation service datasets using four backbone recommendation models to demonstrate the effectiveness of IR-UOF in mitigating UOF and improving recommendation fairness. Furthermore, we select two general digital service datasets to prove that IR-UOF can be extended to tackle the general SOF issue in other types of digital service networks. In summary, the IR-UOF framework achieves optimal model performance across all datasets, while improving fairness by at least 3.8% in recommendation systems and 24.7% in general service systems.
- 31 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Services Computing
- Vol. 18 (5) , 3052-3064
Multi-Objective Unlearning in Recommender Systems via Preference Guided Pareto Exploration
Recommender systems typically collect and analyze user data, which raises the risk of privacy invasion. User-sensitive information can be leaked from the user portrait, e.g., user embedding, within recommender models. Therefore, the task of recommendation unlearning has been widely studied, aiming to eliminate the influence of target data on recommender models. This paper explores the extended concept of unlearning, which seeks to remove sensitive user information while retaining the essential information for recommendation purposes. Previous studies have primarily focused on extended unlearning in isolation, e.g., attribute unlearning. However, users often need to fulfill multiple unlearning objectives simultaneously. Therefore, we bridge this gap by introducing post-training multi-objective unlearning, which allows the concurrent fulfillment of multiple unlearning objectives while preserving recommendation performance. Note that the objectives may conflict with each other, leading to the compromise of one objective when minimizing the overall objective value. To address this challenge, we introduce a Pareto exploration approach that incorporates the recommendation performance as optimization guidance, allowing us to obtain the Pareto optimal solution through the trade-off between conflicting objectives. To adapt to practical scenarios where data is not accessible post-training, we utilize a data-free regularization to guide recommendation performance. We conducted extensive experiments on three real-world datasets, which demonstrate the effectiveness of our proposed method.
- 31 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Services Computing
- Vol. 18 (5) , 3052-3064
Multi-Objective Unlearning in Recommender Systems via Preference Guided Pareto Exploration
Recommender systems typically collect and analyze user data, which raises the risk of privacy invasion. User-sensitive information can be leaked from the user portrait, e.g., user embedding, within recommender models. Therefore, the task of recommendation unlearning has been widely studied, aiming to eliminate the influence of target data on recommender models. This paper explores the extended concept of unlearning, which seeks to remove sensitive user information while retaining the essential information for recommendation purposes. Previous studies have primarily focused on extended unlearning in isolation, e.g., attribute unlearning. However, users often need to fulfill multiple unlearning objectives simultaneously. Therefore, we bridge this gap by introducing post-training multi-objective unlearning, which allows the concurrent fulfillment of multiple unlearning objectives while preserving recommendation performance. Note that the objectives may conflict with each other, leading to the compromise of one objective when minimizing the overall objective value. To address this challenge, we introduce a Pareto exploration approach that incorporates the recommendation performance as optimization guidance, allowing us to obtain the Pareto optimal solution through the trade-off between conflicting objectives. To adapt to practical scenarios where data is not accessible post-training, we utilize a data-free regularization to guide recommendation performance. We conducted extensive experiments on three real-world datasets, which demonstrate the effectiveness of our proposed method.
- Open Access
- 10 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 903-913
Hypergraph Convolutional Network for User-Oriented Fairness in Recommender Systems
The service system involves multiple stakeholders, making it crucial to ensure fairness. In this paper, we take the example of a typical service system, the recommender system, to investigate how to identify and tackle fairness issues within the service system. Recommender systems often exhibit bias towards a small user group, resulting in pronounced unfairness in recommendation performance, specifically the User-Oriented Fairness (UOF) issue. Existing research on UOF faces limitations in addressing two pivotal challenges: CH1: Current methods fall short in addressing the root cause of the UOF issue, stemming from an unfair training process between advantaged and disadvantaged users. CH2: Current methods struggle to unveil compelling correlations among users in sparse datasets. In this paper, we propose a novel Hypergraph Convolutional Network for User-Oriented Fairness, namely HyperUOF, to address the aforementioned challenges. HyperUOF serves as a versatile framework applicable to various backbone recommendation models for achieving UOF. To address CH1, HyperUOF employs an in-processing method that enhances the training process of disadvantaged users during model training. To addressCH2, HyperUOF incorporates a hypergraph-based approach, proven effective in sparse datasets, to explore high-order correlations among users. We conduct extensive experiments on three real-world datasets based on four backbone recommendation models to prove the effectiveness of our proposed HyperUOF.
- 10 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 903-913
Open AccessHypergraph Convolutional Network for User-Oriented Fairness in Recommender Systems
The service system involves multiple stakeholders, making it crucial to ensure fairness. In this paper, we take the example of a typical service system, the recommender system, to investigate how to identify and tackle fairness issues within the service system. Recommender systems often exhibit bias towards a small user group, resulting in pronounced unfairness in recommendation performance, specifically the User-Oriented Fairness (UOF) issue. Existing research on UOF faces limitations in addressing two pivotal challenges: CH1: Current methods fall short in addressing the root cause of the UOF issue, stemming from an unfair training process between advantaged and disadvantaged users. CH2: Current methods struggle to unveil compelling correlations among users in sparse datasets. In this paper, we propose a novel Hypergraph Convolutional Network for User-Oriented Fairness, namely HyperUOF, to address the aforementioned challenges. HyperUOF serves as a versatile framework applicable to various backbone recommendation models for achieving UOF. To address CH1, HyperUOF employs an in-processing method that enhances the training process of disadvantaged users during model training. To addressCH2, HyperUOF incorporates a hypergraph-based approach, proven effective in sparse datasets, to explore high-order correlations among users. We conduct extensive experiments on three real-world datasets based on four backbone recommendation models to prove the effectiveness of our proposed HyperUOF.
- Open Access
- 8 January 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on the Web
- Vol. 18 (2) , 1-24
Heterogeneous Information Crossing on Graphs for Session-Based Recommender Systems
Recommender systems are fundamental information filtering techniques to recommend content or items that meet users’ personalities and potential needs. As a crucial solution to address the difficulty of user identification and unavailability of historical information, session-based recommender systems provide recommendation services that only rely on users’ behaviors in the current session. However, most existing studies are not well-designed for modeling heterogeneous user behaviors and capturing the relationships between them in practical scenarios. To fill this gap, in this article, we propose a novel graph-based method, namely H eterogeneous I nformation C rossing on G raphs (HICG). HICG utilizes multiple types of user behaviors in the sessions to construct heterogeneous graphs, and captures users’ current interests with their long-term preferences by effectively crossing the heterogeneous information on the graphs. In addition, we also propose an enhanced version, named HICG-CL, which incorporates the contrastive learning (CL) technique to enhance item representation ability. By utilizing the item co-occurrence relationships across different sessions, HICG-CL improves the recommendation performance of HICG. We conduct extensive experiments on three real-world recommendation datasets, and the results verify that (i) HICG achieves state-of-the-art performance by utilizing multiple types of behaviors on the heterogeneous graph. (ii) HICG-CL further significantly improves the recommendation performance of HICG by the proposed contrastive learning module.
- 8 January 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on the Web
- Vol. 18 (2) , 1-24
Open AccessHeterogeneous Information Crossing on Graphs for Session-Based Recommender Systems
Recommender systems are fundamental information filtering techniques to recommend content or items that meet users’ personalities and potential needs. As a crucial solution to address the difficulty of user identification and unavailability of historical information, session-based recommender systems provide recommendation services that only rely on users’ behaviors in the current session. However, most existing studies are not well-designed for modeling heterogeneous user behaviors and capturing the relationships between them in practical scenarios. To fill this gap, in this article, we propose a novel graph-based method, namely H eterogeneous I nformation C rossing on G raphs (HICG). HICG utilizes multiple types of user behaviors in the sessions to construct heterogeneous graphs, and captures users’ current interests with their long-term preferences by effectively crossing the heterogeneous information on the graphs. In addition, we also propose an enhanced version, named HICG-CL, which incorporates the contrastive learning (CL) technique to enhance item representation ability. By utilizing the item co-occurrence relationships across different sessions, HICG-CL improves the recommendation performance of HICG. We conduct extensive experiments on three real-world recommendation datasets, and the results verify that (i) HICG achieves state-of-the-art performance by utilizing multiple types of behaviors on the heterogeneous graph. (ii) HICG-CL further significantly improves the recommendation performance of HICG by the proposed contrastive learning module.
- Open Access
- 26 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 6190-6201
In-processing User Constrained Dominant Sets for User-Oriented Fairness in Recommender Systems
Recommender systems are typically biased toward a small group of users, leading to severe unfairness in recommendation performance, i.e., User-Oriented Fairness (UOF) issue. The existing research on UOF is limited and fails to deal with the root cause of the UOF issue: the learning process between advantaged and disadvantaged users is unfair. To tackle this issue, we propose an In-processing User Constrained Dominant Sets (In-UCDS) framework, which is a general framework that can be applied to any backbone recommendation model to achieve user-oriented fairness. We split In-UCDS into two stages, i.e., the UCDS modeling stage and the in-processing training stage. In the In-UCDS modeling stage, for each disadvantaged user, we extract a constrained dominant set (a user cluster) containing some advantaged users that are similar to it. In the in-processing training stage, we move the representations of disadvantaged users closer to their corresponding cluster by calculating a fairness loss. By combining the fairness loss with the original backbone model loss, we address the UOF issue and maintain the overall recommendation performance simultaneously. Comprehensive experiments on three real-world datasets demonstrate that In-UCDS outperforms the state-of-the-art methods, leading to a fairer model with better overall recommendation performance.
- 26 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 6190-6201
Open AccessIn-processing User Constrained Dominant Sets for User-Oriented Fairness in Recommender Systems
Recommender systems are typically biased toward a small group of users, leading to severe unfairness in recommendation performance, i.e., User-Oriented Fairness (UOF) issue. The existing research on UOF is limited and fails to deal with the root cause of the UOF issue: the learning process between advantaged and disadvantaged users is unfair. To tackle this issue, we propose an In-processing User Constrained Dominant Sets (In-UCDS) framework, which is a general framework that can be applied to any backbone recommendation model to achieve user-oriented fairness. We split In-UCDS into two stages, i.e., the UCDS modeling stage and the in-processing training stage. In the In-UCDS modeling stage, for each disadvantaged user, we extract a constrained dominant set (a user cluster) containing some advantaged users that are similar to it. In the in-processing training stage, we move the representations of disadvantaged users closer to their corresponding cluster by calculating a fairness loss. By combining the fairness loss with the original backbone model loss, we address the UOF issue and maintain the overall recommendation performance simultaneously. Comprehensive experiments on three real-world datasets demonstrate that In-UCDS outperforms the state-of-the-art methods, leading to a fairer model with better overall recommendation performance.
- Open Access
- 26 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 984-994
Making Users Indistinguishable: Attribute-wise Unlearning in Recommender Systems
With the growing privacy concerns in recommender systems, recommendation unlearning, i.e., forgetting the impact of specific learned targets, is getting increasing attention. Existing studies predominantly use training data, i.e., model inputs, as the unlearning target. However, we find that attackers can extract private information, i.e., gender, race, and age, from a trained model even if it has not been explicitly encountered during training. We name this unseen information as attribute and treat it as the unlearning target. To protect the sensitive attribute of users, Attribute Unlearning (AU) aims to degrade attacking performance and make target attributes indistinguishable. In this paper, we focus on a strict but practical setting of AU, namely Post-Training Attribute Unlearning (PoT-AU), where unlearning can only be performed after the training of the recommendation model is completed. To address the PoT-AU problem in recommender systems, we design a two-component loss function that consists of i) distinguishability loss: making attribute labels indistinguishable from attackers, and ii) regularization loss: preventing drastic changes in the model that result in a negative impact on recommendation performance. Specifically, we investigate two types of distinguishability measurements, i.e., user-to-user and distribution-to-distribution. We use the stochastic gradient descent algorithm to optimize our proposed loss. Extensive experiments on three real-world datasets demonstrate the effectiveness of our proposed methods.
- 26 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 984-994
Open AccessMaking Users Indistinguishable: Attribute-wise Unlearning in Recommender Systems
With the growing privacy concerns in recommender systems, recommendation unlearning, i.e., forgetting the impact of specific learned targets, is getting increasing attention. Existing studies predominantly use training data, i.e., model inputs, as the unlearning target. However, we find that attackers can extract private information, i.e., gender, race, and age, from a trained model even if it has not been explicitly encountered during training. We name this unseen information as attribute and treat it as the unlearning target. To protect the sensitive attribute of users, Attribute Unlearning (AU) aims to degrade attacking performance and make target attributes indistinguishable. In this paper, we focus on a strict but practical setting of AU, namely Post-Training Attribute Unlearning (PoT-AU), where unlearning can only be performed after the training of the recommendation model is completed. To address the PoT-AU problem in recommender systems, we design a two-component loss function that consists of i) distinguishability loss: making attribute labels indistinguishable from attackers, and ii) regularization loss: preventing drastic changes in the model that result in a negative impact on recommendation performance. Specifically, we investigate two types of distinguishability measurements, i.e., user-to-user and distribution-to-distribution. We use the stochastic gradient descent algorithm to optimize our proposed loss. Extensive experiments on three real-world datasets demonstrate the effectiveness of our proposed methods.
- Open Access
- 30 April 2023
- Published by Association for Computing Machinery (ACM)
- p. 449-459
Intra and Inter Domain HyperGraph Convolutional Network for Cross-Domain Recommendation
Cross-Domain Recommendation (CDR) aims to solve the data sparsity problem by integrating the strengths of different domains. Though researchers have proposed various CDR methods to effectively transfer knowledge across domains, they fail to address the following key issues, i.e., (1) they cannot model high-order correlations among users and items in every single domain to obtain more accurate representations; (2) they cannot model the correlations among items across different domains. To tackle the above issues, we propose a novel Intra and Inter Domain HyperGraph Convolutional Network (II-HGCN) framework, which includes two main layers in the modeling process, i.e., the intra-domain layer and the inter-domain layer. In the intra-domain layer, we design a user hypergraph and an item hypergraph to model high-order correlations inside every single domain. Thus we can address the data sparsity problem better and learn high-quality representations of users and items. In the inter-domain layer, we propose an inter-domain hypergraph structure to explore correlations among items from different domains based on their interactions with common users. Therefore we can not only transfer the knowledge of users but also combine embeddings of items across domains. Comprehensive experiments on three widely used benchmark datasets demonstrate that II-HGCN outperforms other state-of-the-art methods, especially when datasets are extremely sparse.
- 30 April 2023
- Published by Association for Computing Machinery (ACM)
- p. 449-459
Open AccessIntra and Inter Domain HyperGraph Convolutional Network for Cross-Domain Recommendation
Cross-Domain Recommendation (CDR) aims to solve the data sparsity problem by integrating the strengths of different domains. Though researchers have proposed various CDR methods to effectively transfer knowledge across domains, they fail to address the following key issues, i.e., (1) they cannot model high-order correlations among users and items in every single domain to obtain more accurate representations; (2) they cannot model the correlations among items across different domains. To tackle the above issues, we propose a novel Intra and Inter Domain HyperGraph Convolutional Network (II-HGCN) framework, which includes two main layers in the modeling process, i.e., the intra-domain layer and the inter-domain layer. In the intra-domain layer, we design a user hypergraph and an item hypergraph to model high-order correlations inside every single domain. Thus we can address the data sparsity problem better and learn high-quality representations of users and items. In the inter-domain layer, we propose an inter-domain hypergraph structure to explore correlations among items from different domains based on their interactions with common users. Therefore we can not only transfer the knowledge of users but also combine embeddings of items across domains. Comprehensive experiments on three widely used benchmark datasets demonstrate that II-HGCN outperforms other state-of-the-art methods, especially when datasets are extremely sparse.