How Much Is Too Much? Leveraging Ads Audience Estimation to Evaluate Public Profile Uniqueness

Abstract

Abstract. This paper addresses the important goal of quantifying the threat of linking external records to public Online Social Networks (OSN) user profiles, by providing a method to estimate the uniqueness of such profiles and by studying the amount of information carried by public profile attributes. Our first contribution is to leverage the Ads audience estimation platform of a major OSN to compute the information surprisal (IS) based uniqueness of public profiles, independently from the used profiles dataset. Then, we measure the quantity of information carried by the revealed attributes and evaluate the impact of the public release of selected combinations of these attributes on the potential to identify user profiles. Our measurement results, based on an unbiased sample of more than 400 thousand Facebook public profiles, show that, when disclosed in such profiles, current city has the highest individual attribute potential for unique identification and the combination of gender, current city and age can identify close to 55% of users to within a group of 20 and uniquely identify around 18% of users. We envisage the use of our methodology to assist both OSNs in designing better anonymization strategies when releasing user records and users to evaluate the potential for external parties to uniquely identify their public profiles and hence make it easier to link them with other data sources.

Information leakage through mobile analytics services

“…For instance, analytics should produce reports only when the audience is composed of a minimum of k significantly different profiles similar to what is done in the Facebook Audience estimation platform [5]. On the other hand, addressing the identity spoofing vulnerability may be more challenging in practice.…”

Touch and You’re Trapp(ck)ed: Quantifying the Uniqueness of Touch Gestures for Tracking

“…-With the same dataset, we measured the amount of information contained in samples from the same gesture and from multiple gestures. 4 We found that 50 features in a single handwriting sample contribute 68.71% of information about users, which increases to 73.7% with multiple samples. We further identified that two or three different gestures combined together reveal more information about users.…”

“…To the best of our knowledge, this is the first study considering the potential of touch gestures to profile users. Our work complements research on other forms of tracking such as through web browsers, host devices, and online social profiles by fingerprinting browser features, device configurations, and user attributes [4,7,13,18,19,24,25,27,38]. -We develop an analytical framework that measures the amount of identifying information (in bits and relative mutual information) contained in touch gestures, represented as feature vectors, at different levels of granularity.…”

“…In Section 4, we discuss the results on the amount of information conveyed by the users of our dataset for different touch gestures and their combinations . We discuss the limitation and future directions 4 We use the relative mutual information as our metric for identifying information. For details, see Section 3. of this work in Section 5.…”

Paying for Likes?

“…To draw a baseline comparison, we also collected page like counts from a random set of 2000 Facebook users, extracted from an unbiased sample of Facebook user population. The original sample was crawled for another project[9], obtained by randomly sampling Facebook public directory which lists all the IDs of searchable profiles.…”