Personal details of Facebook Likes Cambridge Study Revealed
Researchers at the University of Cambridge Psychometrics Center and Microsoft Research Cambridge analyzed a dataset of more than 58,000 Facebook users who have offered their tastes and were able to determine the intimate details and the millions of personality traits. The study raised important questions about online privacy and personalized advertising.
Using only like Facebook, researchers have created statistical models to predict personal details such as race, political affiliation, religion, relationship status, sexual preference and drug addiction. For a person like Facebook alone, the percentage of accuracy varies 65-95%. For example, researchers have 95% accuracy when it came to determining whether a person was Caucasian or African American. They were 88% accurate whether the person was a homosexual or heterosexual and 82% accurate if the person is Christian or Islamic.
Based on the accuracy of their study, the researchers believe that Facebook likes shockingly alone could reveal personal portraits of millions of users worldwide.
Michal Kosinski, who led the research with colleagues from Cambridge and Microsoft Research states "similar predictions could be made from all kinds of digital data, with this kind of inference secondary made with a precision remarkable - predict statistically sensitive information people might not want to reveal given the variety of digital traces people leave behind, it becomes increasingly difficult for people to control "
This.. means that personal information may be collected from search queries, browsing and buying histories, not just Facebook. The result of this study is another example that anything that happens online is far from private. Even if you use privacy settings on social media, your online behavior (research and purchase history) is all that is necessary for advertisers and others to establish who you are.
0 Komentar