Master Project Presentation — Dec 15 2015, 11am SIE Library
AN ONTOLOGY-BASED APPROACH FOR ACTIVITY RECOGNITION FROM SENSORS IN OFFICES
Project Advisor: Dr. Silvia Nittel
Co-Advisor: Dr. Torsten Hahmann
Understanding the behavior of people in a building can be useful to save energy in buildings and improve efficiency of employees. To study the behavior of people in office buildings, sensors can be used to collect raw data about people’s presence in rooms and their movement through buildings. However, to understand higher-level behavioral patterns, we first developed an application ontology, written in Common Logic, that describes this information of interest at a higher level of abstraction. For example, which characteristics constitute a meeting? How are they related to movement data? Following, we translate the defined ontological concepts into MySQL scripts and Java algorithms to ‘mine’ the low-level sensor data stored in a relational database system. With the programs, we can retrieve some important information about offices such as: how frequently are offices and shares spaces utilized? How many meetings take place on average during a week? A number of use cases demonstrate the capabilities of the system.