SocIntLab researcher Gahl Silverman led a presentation and discussion of early research results on categorizing Unintentional Information Leakage (UIL) in social networks at the 2014 Workshop on Statistical Challenges in E-Commerce Research. The Workshop, held in Tel-Aviv Israel, was co-chaired by SocIntLab deputy director Dr. Inbal Yahav.
Security leaks are a main concern of many organizations in today's era of social networks. Organizations have limited control on their employees' activity in public networks, and even less than that, on their friends' and relatives' activities. The latter, on the other hand, have sometimes limited understanding on what public information sharing is appropriate, and what is not.
In this work we focus on a tiny, yet extremely important niche form of information leakage: leakage through comments on Israeli Defense Force (IDF)-related news articles (on Facebook). It is common in such news articles to censor names of the military personnel, and replace them by their first name initial (for e.g. 'Colonel G.'). We find that it is also common for readers to post comments that potentially lead to the identification of these military personnel, and thereby to de-censor the news articles. An example of a 'leak-enabling comment' is shown in Figure 1 (in Hebrew). The headline of the news article is: "The Navy is satisfied with the appointment Colonel G. as the Commander of Flotilla 13: 'The natural choice, he has a unique character'”. Flotilla 13 being a naval special forces unit, standard military censorship protocol are in place protecting the identity of its commanding officers. The second commenter, as a response writes: "Congrats to G. I know him personally. Good choice, good luck". Given the (readily available) identity of that commenter it only takes a few clicks, to find Colonel G., who is a Facebook friend of this commenter. We denote such comments as Unintentional Information leakage (UIL) comments.