BDI agents: Flexibility, personalization, and adaptation for web-based support systems

Users increasingly value personalized, flexible and interactive forms of support. In this chapter we consider the problem of providing decision support to mobile knowledge workers through a mixed-initiative multi-agent system. We describe an abstract architecture for developing mixed-initiative MASs and identify the main components and underlying ontologies that are required to support automated reasoning, problem solving and adaptation. We propose the BDI model of agency for developing agents to whom goals can be delegated and who can work collaboratively with the users and each other. Users and agents interact in a mixed-initiative mode to establish a common ground and the shared goals to be achieved. Plans as the means to bring about the user's goals, are utilized to enable the refinement and fine-tuning of business processes to better conform to the work style of the individual knowledge worker. Furthermore, we consider possible ways of dealing with exceptional circumstances as they arise in the process of executing plans and how an agent can essentially learn to deal with such situations by interacting with other agents as well as the user who is considered an expert. © 2009 Springer-Verlag Berlin Heidelberg.