When invisible computers begin to support user's daily activities in ubiquitous computing environment, a concept of context-awareness begins to have significant importance as a means to understand what is happening in real world. In this paper, we introduce our research projects on context-awareness. First, we introduce a context inference method that monitors acceleration sensor data attached to a mobile handset and infer user's postures dynamically. Then we introduce a smart home control mechanism named synapse. Synapse exploits the operation history of smart appliances to learn user's habits by a hidden Markov model.