Logical activity recognition for understanding of human activities

Recognizing human activities based on user's contextual information is becoming more important to the success of context-aware pervasive computing. In this paper, we define activities and introduce a six-layer activity model. We also describe our algorithm to perform activity recognition. The key contextual information is location, and other pieces of contextual information are also used in our study. To evaluate our algorithm we collected user data from twenty test users while they performed daily activities. We present a comprehensive experimental evaluation and discuss the results.