{"title":"SBlur: An Obfuscation Approach for Preserving Sensitive Attributes in Recommender System","authors":"Falguni Roy , Na Zhao , Xiaofeng Ding","doi":"10.1016/j.ipm.2025.104282","DOIUrl":null,"url":null,"abstract":"<div><div>User interaction in the recommender system is treated as a way of expressing user preferences, which later serve as input to provide more accurate recommendations. However, such interaction data can be exploited to infer user private attributes, including gender, age, and personality traits, posing significant privacy implications. Existing obfuscation-based approaches endeavor to mitigate these vulnerabilities by adding or removing interactions from user profiles before or during recommender algorithm training. Nevertheless, these methods often compromise recommendation accuracy while facing challenges such as the cold-start user problem and the “rich get richer” effect, undermining recommendation diversity. To address these constraints, we propose SBlur, a strategic obfuscation approach designed to preserve users’ attribute privacy while balancing the privacy-accuracy-fairness trade-off and enhancing diversity. SBlur conceals gender inference attacks by strategically adding and removing items, supported by a combined similarity measure that integrates rating-based and genre preference-based similarities. This combined similarity enables precise user profile personalization for obfuscation, particularly in cold-start scenarios. We evaluate SBlur using three popular datasets (ML100k, ML1M, and Yahoo!Movie) and three state-of-the-art recommendation algorithms (UserKNN, ALS, and BPRMF). Experimental results demonstrate that SBlur achieves a balanced trade-off between privacy, recommendation accuracy, and fairness while promoting recommendation diversity.</div></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":"62 6","pages":"Article 104282"},"PeriodicalIF":6.9000,"publicationDate":"2025-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Information Processing & Management","FirstCategoryId":"94","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0306457325002237","RegionNum":1,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMPUTER SCIENCE, INFORMATION SYSTEMS","Score":null,"Total":0}
引用次数: 0
Abstract
User interaction in the recommender system is treated as a way of expressing user preferences, which later serve as input to provide more accurate recommendations. However, such interaction data can be exploited to infer user private attributes, including gender, age, and personality traits, posing significant privacy implications. Existing obfuscation-based approaches endeavor to mitigate these vulnerabilities by adding or removing interactions from user profiles before or during recommender algorithm training. Nevertheless, these methods often compromise recommendation accuracy while facing challenges such as the cold-start user problem and the “rich get richer” effect, undermining recommendation diversity. To address these constraints, we propose SBlur, a strategic obfuscation approach designed to preserve users’ attribute privacy while balancing the privacy-accuracy-fairness trade-off and enhancing diversity. SBlur conceals gender inference attacks by strategically adding and removing items, supported by a combined similarity measure that integrates rating-based and genre preference-based similarities. This combined similarity enables precise user profile personalization for obfuscation, particularly in cold-start scenarios. We evaluate SBlur using three popular datasets (ML100k, ML1M, and Yahoo!Movie) and three state-of-the-art recommendation algorithms (UserKNN, ALS, and BPRMF). Experimental results demonstrate that SBlur achieves a balanced trade-off between privacy, recommendation accuracy, and fairness while promoting recommendation diversity.
期刊介绍:
Information Processing and Management is dedicated to publishing cutting-edge original research at the convergence of computing and information science. Our scope encompasses theory, methods, and applications across various domains, including advertising, business, health, information science, information technology marketing, and social computing.
We aim to cater to the interests of both primary researchers and practitioners by offering an effective platform for the timely dissemination of advanced and topical issues in this interdisciplinary field. The journal places particular emphasis on original research articles, research survey articles, research method articles, and articles addressing critical applications of research. Join us in advancing knowledge and innovation at the intersection of computing and information science.