A Personal Agent for Combining the Home Appliance Services and Its Learning Mechanism

In this paper, we propose a new personal agent for generating the combinational services from using history of appliances in the home network environment. In such environment, it is required that flexible services can be provided by combining services of appliances and unskillful users can use these services without knowledge. So, it is needed to satisfy following: (1) combinational services can be suggested automatically and (2) the increase of services can be followed. Then, we propose a new personal agent that suggests combinational services by learning the lifestyle. Its learning mechanism is based on Self-Organizing Map (SOM), and can follow the increase of services. We implemented the the agent, and use history of a user for two weeks was made to learn. As the result, we confirmed that the agent can extract services related with time or location and can suggest combinational services.