METHOD AND SYSTEM FOR DETECTING HUMAN ABNORMAL BEHAVIOUR BY USING DATA FROM USER'S PORTABLE DEVICES

摘要

A system and method for detecting human abnormal behaviour, comprising data preprocessing (110) which comprises: defining time intervals (111) for a user by extracting (100) intervals based on historical data from raw sensors (101) and personalizing the intervals based on active data (102) input by the user or inferred data from the raw sensors (101); constructing words (112) for each interval as a vector comprising a plurality of sensor- based feature levels being mapped to a range of values of a sensor-based feature extracted from historical and real raw sensors data (101); constructing a document (113) with the words based on the defined intervals and the mapped sensor-based feature levels. After data preprocessing (110), behaviour routines are extracted (120) for each interval by extracting the most frequent words using topic modelling. If a deviation is detected between the sensor-based features extracted in a current time interval and a past routine extracted for a past time interval previous to the current time interval, an alert is generated (130).