Wednesday 11th June 2014, 3:45pm
TITLE | The Human Manifold: On the Predictability of Human Online Behaviour and its Consequences
ABSTRACT | 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 including: sexual orientation, ethnicity, religious and political views, personality traits, intelligence, happiness, use of addictive substances, parental separation, age, and gender. 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. The proposed model uses dimensionality reduction for pre-processing the Likes data, which are then entered into logistic/linear regression to predict individual psycho-demographic profiles from Likes. The model correctly discriminates between homosexual and heterosexual men in 88% of cases, African Americans and Caucasian Americans in 95% of cases, and between Democrat and Republican in 85% of cases. For the personality trait “Openness,” prediction accuracy is close to the test–retest accuracy of a standard personality test. We give examples of associations between attributes and Likes and discuss implications for online personalization and privacy. This is joint work with Michal Kosinski and David Stillwell at the University of Cambridge and is based on the PNAS paper “Private traits and attributes are predictable from digital records of human behavior”.
BIOGRAPHY | Dr. Thore Graepel is a researcher at Microsoft Research Cambridge leading the Online Services and Advertising and Applied Games group, where their work is focused on the application of large scale machine learning and probabilistic modelling techniques to a wide range of problems including online advertising, web search, and games. He has a particular passion for the game of Go and the quest for developing a Go engine that plays as well as the best human players. More recently, he has been investigating crowdsourcing, collective intelligence and social networking data. Before joining the Cambridge lab of Microsoft Research, Thore was a postdoctoral researcher at the Department of Computer Science at Royal Holloway, University of London working on learning theory and machine learning algorithms with Prof. John Shawe-Taylor. Previous to that, he worked with Nici Schraudolph and Prof. Petros Koumoutsakos as a postdoctoral researcher at the Institute of Computational Science (ICOS) which is part of the Department of Computer Science of the Swiss Federal Institute of Technology, Zürich (ETH). Topics of research were machine learning and large-scale nonlinear optimisation. Dr. Graepel received his doctorate (Dr. rer. nat) from the Department of Computer Science of the Technical University of Berlin, where he was first a member of the Neural Information Processing group of Prof. Klaus Obermayer and later joined the Statistics group of Prof. Ulrich Kockelkorn.