Berlin 2018 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

SOE: Fachverband Physik sozio-ökonomischer Systeme

SOE 4: Social Systems, Opinion and Group Dynamics I

SOE 4.1: Talk

Monday, March 12, 2018, 11:30–11:45, MA 001

Predicting hidden user properties using online egocentric networks — •Gábor Tamás and János Török — Department of Theoretical Physics, Budapest University of Technology and Economics, H-1111 Budapest, Hungary

Social physics applies the methods of statistical physics to study the effects of human relations. Social networks are considered to be composed of humans as nodes and social relations as links between them. Using data from online social networks it is possible to study the properties and behavior of humans and even pin out and correct errors and missing information. Here we aim to predict the age of egos with a very simple method. This is an important issue in a world when privacy is a key issue.

We had access to a Hungarian social network site, where the connections and the birth date of the registered users were given. The basis of our method is to use the egocentric network of people to determine its communities and the average age within them. We found that most of our acquaintances have similar age as we have or they are 25 year younger or older (different generation) which could be separated by histogram technique.

Our algorithm does not use machine learning methods and is based only on a few assumptions. It was very efficient, in some cases the code predicted the age of the ego with more than 90 % probability with 2 years accuracy. Our success suggests that we need the privacy of our friends to hide our properties.