Article
Computer Science, Information Systems
Xin Xin, Jiyuan Yang, Hanbing Wang, Jun Ma, Pengjie Ren, Hengliang Luo, Xinlei Shi, Zhumin Chen, Zhaochun Ren
Summary: Modern recommender systems predict users' future interactions based on their historical behavior data, and generate exposure data to provide personalized recommendations. However, the privacy-sensitive user behavior data can be inferred through the modeling of system exposure, posing a high risk of leakage. An attack model is proposed to infer user's historical behavior from the current system exposure. Experimental results demonstrate the danger of user behavior data leakage. To address this, a two-stage privacy-protection mechanism is proposed, which shows a trade-off between recommendation accuracy and privacy disclosure risk.
ACM TRANSACTIONS ON INFORMATION SYSTEMS
(2023)
Article
Computer Science, Theory & Methods
Bo Jiang, Mohamed Seif, Ravi Tandon, Ming Li
Summary: This paper addresses the problem of answering count queries for genomic data under perfect privacy constraints. The authors propose local and central count-query mechanisms that achieve perfect information-theoretic privacy while minimizing the expected absolute error. They also derive a lower bound for the per-user probability of error and show that their mechanisms achieve error close to this bound. Numerical experiments demonstrate that the performance of each mechanism depends on the data prior distribution, the intersection between queried and sensitive genotypes, and the correlation strength in the genomic data sequence.
IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY
(2023)
Article
Computer Science, Artificial Intelligence
Zongda Wu, Shigen Shen, Haiping Zhou, Huxiong Li, Chenglang Lu, Dongdong Zou
Summary: This paper proposes a method to protect users' commodity viewing privacy by constructing dummy requests on a trusted client to confuse and cover up user preferences on the untrusted server-side. The study introduces a privacy model and an implementation algorithm to measure the effectiveness of confusion and cover-up effects. The results show that the proposed approach effectively enhances the security of users' commodity viewing privacy on the untrusted server-side.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Information Science & Library Science
Fei Bu, Nengmin Wang, Bin Jiang, Qi Jiang
Summary: This study examines the determinants of Privacy by Design (PbD) implementation from the perspective of information system (IS) engineers, finding that the engineers' attitude significantly impacts both their behavioral intention and implementation behavior. The research highlights the importance of IS engineers' attitude towards PbD usage as a key factor for successful PbD implementation.
INTERNATIONAL JOURNAL OF INFORMATION MANAGEMENT
(2021)
Article
Business
Jin Li, Yulan Zhang, Jian Mou
Summary: As a new social media platform, short-form video apps combine e-commerce and social networking functions. This study aims to understand the impact mechanism of private information disclosure behaviors and benefits obtained on privacy concerns of short-form video platform users. Through analyzing the data of 35,456 users on TikTok, it was found that user behaviors and privacy sensitivity had a correlation, with popularity playing a mediating role. The official certification from the platform moderated the impact of popularity on privacy sensitivity. This study enriches the understanding of the correlation between privacy concerns and information disclosure behavior, and provides practical implications for privacy settings and protections by short-form video platforms.
JOURNAL OF RETAILING AND CONSUMER SERVICES
(2023)
Review
Medicine, General & Internal
Brooke Rockwern, Dejaih Johnson, Lois Snyder Sulmasy
Summary: The position paper discusses the importance of technological advancements and privacy protection in the digital health landscape. It presents a series of policy principles and recommendations to strengthen the privacy protection of medical information and uphold individual rights and a culture of trust.
ANNALS OF INTERNAL MEDICINE
(2021)
Article
Computer Science, Information Systems
Y. A. Nanehkaran, Zhu Licai, Junde Chen, Qiu Zhongpan, Yuan Xiaofeng, Yahya Dorostkar Navaei, Sajad Einy
Summary: In this paper, a medical recommender system is presented for identifying and treating chronic diseases using an IoT device. The K-nearest neighbor classification method is used to identify the disease type, while collaborative filtering is used to recommend appropriate treatments. The results show that this approach has higher accuracy in diagnosing and predicting chronic diseases compared to previous methods.
WIRELESS COMMUNICATIONS & MOBILE COMPUTING
(2022)
Article
Chemistry, Multidisciplinary
Zhong Zhang, Minho Shin
Summary: In the realm of mobile privacy, there are various attack methods to leak users' private information. Despite protection mechanisms against privilege escalation, attackers can utilize inference algorithms to derive new information or enhance data quality without violating privilege limits. A proposed detection and protection mechanism using Inference Graph and Policy Engine allows users to control their privilege policies in information escalation, showing feasibility and good usability in implementation results.
APPLIED SCIENCES-BASEL
(2021)
Article
Computer Science, Information Systems
Wei Du, Ang Li, Pan Zhou, Ben Niu, Dapeng Wu
Summary: Large volumes of video data recorded by mobile devices and embedded sensors can be used to answer queries about our lives, the physical world, and our evolving society. However, running convolutional neural networks (CNNs) directly on these devices can be burdensome due to their limited capacity. To address this issue, we propose a privacy-preserving and computationally efficient framework for mobile video analytics.
IEEE TRANSACTIONS ON MOBILE COMPUTING
(2022)
Article
Computer Science, Artificial Intelligence
Ying Hu, Yong Zhang, Xiaozhi Gao, Dunwei Gong, Xianfang Song, Yinan Guo, Jun Wang
Summary: Feature selection is an important preprocessing technique in data mining and machine learning. This paper proposes a federated feature selection framework that introduces a trusted third participant to process and integrate optimal feature subsets from multiple participants. A federated evolutionary feature selection algorithm based on particle swarm optimization is proposed to effectively solve feature selection problems with multiple participants under privacy protection. Experimental results show that the proposed algorithm can significantly improve the classification accuracy of the feature subset selected by each participant while protecting data privacy.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Environmental Sciences
Ming Zhuo, Wen Huang, Leyuan Liu, Shijie Zhou, Zhiwen Tian
Summary: Space Information Networks, represented by satellite internet, are developing rapidly. However, accurate information on networks cannot be shared due to safety or confidentiality reasons. To address this issue, a differentially private mechanism is designed to protect the privacy of network entities while sharing statistical information.
Article
Computer Science, Artificial Intelligence
Siwei Feng
Summary: This paper proposes a VFL-based feature selection method that utilizes deep learning models and complementary information from multiple party samples without data sharing. Extensive experiments demonstrate the effectiveness of this approach in collaborative feature selection without data disclosure.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Multidisciplinary Sciences
Jayakrishnan Ajayakumar, Andrew Curtis, Jacqueline Curtis
Summary: This paper examines the use of Zip4 codes as a data layer for sharing fine-scale spatial health data, which enables privacy preservation while maintaining suitable analytical precision. The results suggest that Zip4 data provides more insightful geographic output compared to other commonly used aggregation units. However, the use of Zip4 codes also poses a risk to spatial anonymity, and researchers and analysts need to be aware of the potential confidentiality violations.
Article
Computer Science, Information Systems
Zongda Wu, Huawen Liu, Jian Xie, Guandong Xu, Gang Li, Chenglang Lu
Summary: This paper proposes an agent-based algorithm for the protection of user privacy health topics based on identity replacement. It also introduces a client-based algorithm for the selection of intermediate agents to improve security. The effectiveness of the proposed method is demonstrated through theoretical analysis and experimental evaluation.
WORLD WIDE WEB-INTERNET AND WEB INFORMATION SYSTEMS
(2023)
Article
Green & Sustainable Science & Technology
Shuhuan Zhou, Yajuan Liu
Summary: This study examined the effects of perceived privacy risks and benefits on the online privacy protection behaviors of Chinese teens. The data was collected from teens in seven provinces of Mainland China and analyzed using a structural equation model (SEM). The study found that perceived privacy benefits had no significant impact on information privacy concerns and online privacy protection behaviors, while perceived privacy risk had a significantly positive effect on online privacy protection behaviors. Additionally, information privacy concerns mediated the effects of perceived privacy risk on Chinese teens' online privacy protection behaviors.
Article
Computer Science, Information Systems
Qian Li, Xiangmeng Wang, Zhichao Wang, Guandong Xu
Summary: In recommendation systems, the MNAR problem causes selection bias and affects recommendation performance. Existing approaches focus on modeling the exposure of missing entries but overlook the causal perspective. To address this, we propose a method called DENC (De-Bias Network Confounding in Recommendation) that provides causal analysis on MNAR using both inherent factors and auxiliary networks. DENC includes an exposure model to control confounding and a deconfounding model for balanced representation learning. Experiments on three datasets demonstrate that DENC outperforms state-of-the-art baselines.
ACM TRANSACTIONS ON KNOWLEDGE DISCOVERY FROM DATA
(2023)
Article
Computer Science, Artificial Intelligence
Yicong Li, Hongxu Chen, Yile Li, Lin Li, Philip S. Yu, Guandong Xu
Summary: Path exploration is an effective method for improving the performance of explainable recommendation systems. However, existing methods have some limitations, such as ignoring dynamic user-item evolutions and low-quality paths. In this study, we propose a novel Temporal Meta-path Guided Explainable Recommendation leveraging Reinforcement Learning (TMER-RL) method, which uses attention mechanisms and reinforcement learning to model dynamic user-item evolutions and enhance the quality and explainability of paths.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Computer Science, Information Systems
Abdul Rehman, Imran Razzak, Guandong Xu
Summary: Federated learning is a promising framework for preserving user data privacy in machine and deep learning models. This study proposes a novel framework based on federated learning to detect side-channel attacks in smartphone keyboard input, and experimental results demonstrate its effectiveness.
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
(2023)
Article
Computer Science, Artificial Intelligence
Zongda Wu, Jian Xie, Shigen Shen, Chongze Lin, Guandong Xu, Enhong Chen
Summary: In this article, a privacy protection solution is proposed for a Chinese keyword-based book search service. By modifying the user query sequence, the user query topics can be confused to protect the user's topic privacy on an untrusted server, while maintaining the accuracy of the book search service. A client-based framework and privacy model are introduced to guide the generation of cover queries based on a user query sequence. The proposed modification algorithm quickly generates cover queries that meet the privacy model by replacing, deleting, and adding keywords. The effectiveness of the approach is demonstrated through theoretical analysis and experimental evaluation, showing improved security for users' topic privacy without compromising efficiency, accuracy, and usability of the existing Chinese keyword book search service, thus contributing to the construction of a privacy-preserving text retrieval platform in an untrusted network environment.
ACM TRANSACTIONS ON ASIAN AND LOW-RESOURCE LANGUAGE INFORMATION PROCESSING
(2023)
Editorial Material
Computer Science, Information Systems
Imran Razzak, Muhammad Khuram Khan, Guandong Xu, Fahmi Khalifa
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
(2023)
Editorial Material
Computer Science, Information Systems
Hongzhi Yin, Yizhou Sun, Guandong Xu, Evangelos Kanoulas
ACM TRANSACTIONS ON INFORMATION SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Tri Dung Duong, Qian Li, Guandong Xu
Summary: Counterfactual fairness addresses discrimination between model predictions in the actual and counterfactual worlds. This research proposes a minimax game-theoretic model that achieves counterfactual fairness without strict assumptions on structural causal models.
EXPERT SYSTEMS WITH APPLICATIONS
(2024)
Article
Computer Science, Artificial Intelligence
Xiangmeng Wang, Qian Li, Dianer Yu, Peng Cui, Zhichao Wang, Guandong Xu
Summary: This study proposes an unbiased and semantic-aware disentanglement learning approach from a causal perspective, which generates semantic-aware representations by disentangling users' true intents from specific item context, and designs a causal intervention mechanism to eliminate confounding bias.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Computer Science, Information Systems
Haoran Yang, Xiangyu Zhao, Muyang Li, Hongxu Chen, Guandong Xu
Summary: This paper investigates the implementation of differential privacy (DP) on graph edges and observes a decrease in performance. The authors propose leveraging graph contrastive learning to alleviate the performance drop resulting from DP.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Artificial Intelligence
Xingliang Wang, Dongjing Wang, Dongjin Yu, Runze Wu, Qimeng Yang, Shuiguang Deng, Guandong Xu
Summary: This paper proposes an Intent Aware Graph Neural Network-based model (IAGNN) for predicting/recommending the next potential POI that a user may interact with. The model utilizes the user's check-in behavior sequences as graphs and uses a graph neural network to learn the feature representations of POIs. It also incorporates a hierarchical attention network and a user intent-aware module to capture users' preferences and intents. Experimental results show that the proposed IAGNN model outperforms baseline models in terms of recall and MRR.
Article
Computer Science, Artificial Intelligence
Qian Li, Zhichao Wang, Shaowu Liu, Gang Li, Guandong Xu
Summary: Treatment effect estimation is crucial for answering questions about the impact of specific treatments. However, treatment assignment bias is a fundamental issue in this research. To address this problem, the paper proposes a novel causal optimal transport (CausalOT) model that utilizes global information on observational covariates to mitigate limited overlapping, and designs a new counterfactual loss to improve estimation accuracy.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Review
Computer Science, Cybernetics
Li He, Guandong Xu, Shoaib Jameel, Xianzhi Wang, Hongxu Chen
Summary: This article introduces a graph-based model for detecting spam information in product reviews on e-commerce platforms. The model utilizes relevant metadata and relational data to capture semantic information in the review text, and synthesizes interactions at different levels through fusion techniques. Experimental results show that the proposed model outperforms other baselines.
IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS
(2023)
Article
Computer Science, Cybernetics
Zhihong Cui, Xiangguo Sun, Hongxu Chen, Li Pan, Lizhen Cui, Shijun Liu, Guandong Xu
Summary: Nowadays, dynamic recommendations struggle to find user online interest evolving patterns due to their complex interactions. Each interaction is influenced by multiple underlying reasons, so it is necessary to analyze each interaction instance individually. Additionally, traditional sequential models fail to personalize recommendations for users with different long-term and short-term tastes. In this article, a novel recommendation model based on Graph Diffusion and Ebbinghaus Curve is proposed, which explores underlying reasons for interactions and captures users' personalized strategies using a well-designed neural network and the Ebbinghaus Curve. Extensive experiments on real-world datasets confirm the superiority of the proposed model over existing baselines.
IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS
(2023)
Article
Computer Science, Hardware & Architecture
Hadeel Alhosaini, Sultan Alharbi, Xianzhi Wang, Guandong Xu
Summary: This paper provides a comprehensive review of API recommendation research for mashup creation. It compares different approaches and techniques, identifies challenges in current research, and suggests promising directions for future studies. It serves as a useful reference for researchers and practitioners in this field.
Article
Automation & Control Systems
Yakun Chen, Ruotong Hu, Zihao Li, Chao Yang, Xianzhi Wang, Guodong Long, Guandong Xu
Summary: This study proposes a novel approach to deal with missing values in multivariate time series. The approach combines graph and recurrent neural networks to capture dependencies between variables and time. By integrating external data sources, such as domain knowledge and implicit relationships between nodes, the ability to impute missing values is enhanced. Experimental results show that the model outperforms current state-of-the-art baselines in various industrial fields.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2024)