A User Activity Recognition Framework for Supporting Electricity Planning in Campus Library

Abstract

To assist on campus library management on electricity usage, activity recognition is applied to detect students’ actions in library environment. An RFID was selected as a tool to identify and detect users’ action. A supervised data-driven method approach is invented to generate an activity rule set in activity recognition. The activity data were used in an automatic process to manage rooms specified by activity based on two main factors, i.e. past activity data and room capacity. From experiment, the results of accuracy indicated that the proposed method got the satisfied results for 0.993 precision and 0.988 recall score in overall. The room management plan using the obtained acidity data yielded the plan to lower the weekly electricity usage for approximately 4,089 units from original setting of 5,962 units which are equivalent to 31.53 percent reduction of electricity.