ECAI-2002 Conference Paper

[full paper]

An Intelligent Inference Approach for User Interaction Modeling in a Generic Agent Based Interface System

Yanguo Jing, Nick Taylor, Keith Brown

Research has shown that a typical user exhibits a certain pattern when interacting a computer system. This work infers the user interaction habits automatically from the actions that the user and the application produce. The user interaction habits can enable an interface agent to provide assistance to a user or operate the application on the user’s behalf. All these actions are automatically segmented into several meaningful action sequences, which are stored in the user’s interaction history. An inference algorithm is used to construct the user’s interaction model from his/her interaction history. Probabilities of each action’s next possible action are calculated. Interface agents then provide assistance based on a user’s interaction model. The interaction model is modified automatically as the interface agents collect more actions. This method is embedded within a generic agent based interface system. Information on the prototype system, some experiments and future work are presented.

Keywords: Interaction Model, Interaction Habits, User Model, Intelligent Interface Agents, Task, Inference

Citation: Yanguo Jing, Nick Taylor, Keith Brown: An Intelligent Inference Approach for User Interaction Modeling in a Generic Agent Based Interface System. In F. van Harmelen (ed.): ECAI2002, Proceedings of the 15th European Conference on Artificial Intelligence, IOS Press, Amsterdam, 2002, pp.103-107.