Reasoning on domain knowledge level in human-computer interaction

Abstract

This paper proposes an innovative approach for dynamically analyzing a user's dialog behavior and inferring a user's domain knowledge level simultaneously that combines neural networks, fuzzy cognitive maps, and fuzzy production rules. Further, this approach supports more cooperative human-computer interaction through dialog adaptation. Furthermore, when the user's knowledge level and problem-solving capability are inferred more accurately, there is more assurance that the system's interaction strategy can match more closely to the user's style. This research implements a neural network for classifying a user's performance pattern using UNIX file security commands. Input and output information that relate to a fuzzy cognitive map and fuzzy production rules are explained.