a thumbs up from facebook like

Modeling Sensitive Traits from Facebook Likes

You may remember a recent post on collaborative filtering and some of the amusing mistakes that these algorithms make. Here is the other side of the story. This study from the University of Cambridge and Microsoft Research found that they can predict a ton of very sensitive private information about you based only on your Facebook Likes – which by the way are publicly available.

We show that easily accessible digital records of behavior, Facebook Likes, can be used to automatically and accurately predict a range of highly sensitive personal attributes… The analysis presented is based on a dataset of over 58,000 volunteers who provided their Facebook Likes, detailed demographic profiles, and the results of several psychometric tests…

They were able to predict things like sexual orientation, political affiliation, intelligence, happiness, and many others. On average, each person had Liked 68 things, which served as the source of the algorithm. But there were also some very diagnostic items that even a single Like said about you. Their prediction accuracies were in the 70-90% range – so there were still some errors but this is remarkable considering it was based only on a single source of publicly available information. Anyone can do it – and they probably are!!!

Image credit: “Facebook like thumb” by Enoc vt used under Public domain

One thought on “Modeling Sensitive Traits from Facebook Likes”

Leave a Reply

Your email address will not be published. Required fields are marked *