Abstract
Development of everyday life service technology requires understanding everyday life behavior. However, it is difficult to recognize meanings of multiply ambiguous behavior from collected sensor data. To tackle this problem, this paper proposes a system for describing and managing human behavior data using not only sensor data but also meaning and context data. This system consists of embedded sensors for measuring a person's behavior, wearable sensors for measuring the person's remarks and statistical analysis software. The authors applied the developed system to two analyses : 1) causality analysis between "cleaning-up" pattern and person's temperament, and 2) causality analysis between power consumption and behavior.