부산대학교 도서관

Electronic subscriptions are widespread and quickly becoming the dominant mode of access for services like music and video streaming, news, and academic articles. Although electronic subscriptions are convenient for users, they reveal a lot of information, ranging from personal preferences to geographic movements. Many users want electronic services, but they also want privacy. Simply anonymizing data doesn’t always protect users’ privacy. Indeed, multiple anonymized datasets released for research purposes, including the AOL search dataset1 and the Netflix Prize dataset,2 have been partially deanonymized through correlation or by understanding of the semantics of the released data.