Mining individual behavior pattern based on significant locations and spatial trajectories

Individual behavior pattern captures human-centric significant places, movement trace and regular routines in the daily life. This paper proposes the BP-Mine framework, which consists of three phases, that is, location extraction, trajectory modeling and behavior pattern mining. Raw WiFi RSS readings and accelerometer sensor data are fused to extract fine grained significant locations in user´s daily life. Then, the locations are modeled as trajectories according to the spatial-temporal relationship between locations. The mining phase utilizes different strategies to discover diverse types of behavior patterns by observing long-term regular routines. Therefore, we can mine valuable knowledge from monitor-dependent people´s life style and regularity, which makes sense to supply the accessment of life activity measurement, the suggestion of healthy life mode, etc., for the domiciliary elderly.