Perceived Versus Actual Predictability of Personal Information in Social Networks

摘要: This paper looks at the problem of privacy in context Online Social Networks (OSNs). In particular, it examines predictability different types personal information based on OSN data and compares to perceptions users about disclosure their information. To this end, a real life dataset is composed. consists Facebook (images, posts likes) 170 people along with replies survey that addresses both information, as well sensitivity Importantly, we evaluate several learning techniques for prediction user attributes data. Our analysis shows respect specific are often incorrect. For instance, appears political beliefs employment status higher than they tend believe. Interestingly, also characterized by more sensitive, actually easily predictable think, vice versa (i.e. relatively less sensitive might have thought).