It is anticipated that electronics will be embedded into the environment around us, and devices will be connected together into meaningful systems, making it possible for us to access context and services wherever and whenever we want with natural and human-centered interaction. For the realization of this concept, devices and applications have to be personalized and aware of their context. When humans talk with humans, they are able to use implicit situational information to increase the conversational bandwidth. Just like this, by improving a computing system's access to context, we may increase the richness of communication in human-computer interaction and make it possible to produce more useful computational services.
Context-aware applications can work by forming interactive systems of necessary sources, and designers need to know what the essential constituents of the systems are and to design the overall structure of effective systems. This research aims to develop a system structure for context-aware applications, which perceive user behavior patterns and produce intelligent services, based on fundamental knowledge of context-awareness and analysis of related research instances. The research procedure had the following stages; Firstly, a knowledge base of context-awareness was formulated by synthesizing preceding researcher's definitions and classification. This knowledge is expected to help designers choose adequate contexts and services when they plan to design a context-aware application. Secondly, some existing context-aware applications and frameworks were analyzed to investigate the present state of related researches and apply useful factors to a new conceptual system structure. Thirdly, a system structure aware of user behavior patterns was developed by deriving solutions of problems inspected by the checklist, which had been created through analysis of design process of context-aware applications. Lastly, on the basis of behavior patterns observed from 6 home-theater users, the developed structure was specified with 4 scenarios about awareness of user behavior patterns.