Anomaly Detection Using Smartphone Sensors for a Bullying Detection

Anomaly Detection is a fundamental process of detecting a situation different from the ordinary. The following work deals with anomalies in the human behavioral domain while filling out a questionnaire about bullying and cyberbullying. In this work, data obtained from smartphones’ sensors (accelerometer, magnetometer, and gyroscope) are analyzed to apply useful Anomaly Detection techniques to detect any abnormal behaviors adopted while filling out the questionnaire implemented in an Android application. Psychology and computer science are merged to analyze and detect any latent patterns within the data set under examination to understand any polarizing content proposed during the use of the app and identify users who exhibit anomalous behaviors, possibly common to classes of users.