Face 2 Facebook: Face-Recognition Database Datingsite
Niklas schreibt mir: „Face 2 Facebook und Lovely Faces von Poalo Cirio und Alessandro Ludovico, die auch schon Google will eat itself und Amazon Noir gemacht haben, ging gestern online. Aus ihrer Mail: ‘Stealing 1 million Facebook profiles, filtering them with face-recognition software, and then posting them on a custom-made dating website, sorted by their facial expressions characteristics.’“ Die haben eine Million Datensätze aus Facebook über eine Datenbank in eine Datingwebsite verwandelt, alles im Namen der Kunst. Schönes Spiel mit Social Network-Mechanismen. Von ihrer „How we did it“-Seite:
Through special custom software we collected data from more than 1,000,000 Facebook users. What we collected is their “public data” – some of their personal data (name, country, Facebook groups they subscribe to) plus their main profile picture and a few friend relationships. We built a database with all this data, then began to analyze the pictures that showed smiling faces. The vast majority of pictures were both amateurish and somehow almost involuntarily or unconsciously alluring. And they are almost always “smiling”. […]
Once the database was ready, we studied and customized a face recognition algorithm. The algorithm used self learning neural networks and was programmed to “group” the huge amount of faces we collected (and their attached data) in a few simple categories. The categories are among the most popular that we usually use to define a person at a distance, without knowing him/her, or judging based only on a few behaviors. We picked six categories (“climber”, “easy going”, “funny”, “mild”, “sly” and “smug” – working definitions), with some intuitive differences, for both male and female subjects. The software effectively extracted 250,000 faces that were connected to the relevant public data in our database. […]
We established a dating website [www.Lovely-Faces.com], importing all the 250,000 profiles. This step builds the virtual land that Facebook is always close to but never explicitly steps in, being just an enormous background to the active process of searching for potential sexual relationships. The profiles will be definitively “single” and available, in a fairly competitive environment, with real data and real faces that users have personally posted. Their smiles will finally reach what they unconsciously really want: more relationships with unknown people, attracted by their virtual presence.