INTECIN   20395
INSTITUTO DE TECNOLOGIAS Y CIENCIAS DE LA INGENIERIA "HILARIO FERNANDEZ LONG"
Unidad Ejecutora - UE
congresos y reuniones científicas
Título:
What do our digital records reveal about us? Automatic Prediction of Demographic and Psychometric Attributes from Digital Data
Autor/es:
MARIANO G. BEIRO; KYRIAKI KALIMERI; CIRO CATTUTO; ROBERT RALEIGH
Lugar:
Colonia
Reunión:
Conferencia; 3rd Annual International Conference on Computational Social Science (IC2S2); 2017
Resumen:
This paper demonstrates how easily accessible digital records of behavior, like webpage history data or mobile application usage, can be used to automatically and accurately infer highly complex personal attributes such as our moral positions and personal values. The analysis presented is based on a large-scale dataset closely representative to the US census. Approximately 7,500 participants accepted to participate in the study providing detailed demographic profiles, psychometric assessments and digital records of web browsing or mobile phone data. The direct involvement of the participants providing their own opinions and views rendered the procedure independent of external annotators or any other kind of proxy data. At the same time, the digital interactions provide fine-grained records of human behavior at individual level, yet are global in scale. The simultaneous availability of accurate survey data combined along with the digital behavioral ones permits the development of a prediction model which accurately infers the former from the latter, providing at the same time insights of the most predictive behaviours that differentiate between moral foundations, human values and a variety of demographic attributes (i.e. gender, political views and others).