A Multi-agent Based Testbed for Agent Interface Evaluation

This study proposes a conceptual model of testbed to support multimodal interaction evaluation throughout the agent-based user interfaces development. Taking into account the Neimo system´s requirements, the conceptual model consists of human test subject, agent-based user interface, evaluation analysis, and application simulaionr. Both evaluation analysis and application simulation cooperate with each other to form the testbed. A multi-agent based architecture is also designed to construct the testbed. The multi-agent based architecture aims to achieve a generic, flexible, and portable testbed to support user observation and analysis in an efficient way, also independent of the implementation of the system application functions.