Towards Personality Detection and Prediction using Smartphone Sensor Data

The emergence and adoption of mobile smartphones and their ubiquitous use by mobile phone users provide a distinctive opportunity for the collection of multi-modal sensor data. The data collected from such smartphone devices can be used to observe, recognize and predict patterns in user behaviors such as physical activity, psycho-physiological conditions such as affective state, personality traits such as extraversion, emotionality, and honesty-humility, as well as other mental health-related or psychological variables. In this work, we design and develop a system that can be used for the unobtrusive collection and storage of anonymized smartphone sensor data for personality detection and prediction. Our proposed system is developed across the two main smartphone operating systems to cover a wide range of users and to consider the available sensors across platform-dependent devices, as well as to account for distinctions in the personality traits of users of different operating systems. We present a correlation analysis and describe how the dataset collected from the smartphone sensors, which were deployed in a large-scale psychology study with about 200 participants, can be used to detect and predict different personality states.