A Multiagent system for the Life Long Personalized Task Coordination based on the user behavior patterns

The core of the ubiquitous network technology is to provide high availabilities of network environments. This implies that it is possible to provide services which are appropriate to user contexts. For the context-aware services, there have been various kinds of service applications, but they are not flexible enough to reflect unexpectedly changing user contexts and dynamic environments. The main reason is that service scenarios are predefined at an early stage-compile time, so they cannot adapt to environmental changes. In this paper, proposed the multiagent system architecture for the life long personalized tasks (LLPT). This architecture enables to continuously reflect the changes of personal behavior patterns by dynamically generating customized service applications for each user during his/her lifetime. In other words, application scenarios do not need to be determined at compile time, since they can be dynamically generated at run time. This flexibility is brought from a life-long learning system which learns user behavior patterns for the system life time and produce behavior pattern rules of each user