Natural Interaction Using the Adaptive Agent-Oriented Software Architecture (AAOSA)

摘要

We need to reverse the paradigm in which we contort the expression of our intent to the dictates of the machines functionality and instead turn functionality into an expression of our intent. In this paper we propose an Agent Oriented design for applications where agents interpret the intent of the user and map that onto the applications functionality. Our approach is based on the Adaptive Agent Oriented Software Architecture (AAOSA), a framework providing support for applications structured as communities of collaborative claim-based and message-driven components. An application in AAOSA is designed as a network of agents, where each agent is responsible for all its data structures and operations and takes part in the decision on how to employ them. Therefore, there will only be one agent, or sub-network of agents, responsible for all processes and data structures related to a certain sub-domain. As a result, to add agents to applications simply extends the functionality with only minimal - if any - disruptions to the original behavior of the application. Furthermore, AAOSA uses contextual information to coordinate and resolve contradicting claims, and will in some cases also rely on emergent behavior to achieve the desired result. Based on this approach, AAOSA makes it possible for applications to better incorporate context into the user interactions, thereby making applications more flexible and easy to use.