Active user interface in a knowledge discovery and retrieval system

We discuss the construction of an active user interface agent in our on-going research towards the development of a system to help users retrieve information and discovery knowledge for a medical domain application. The primary goal of the interface agent is to adaptively react to the dynamic changes in the user's interests and preferences in searching for information within the context of the on-going information retrieval task. The agent dynamically constructs and maintains a decision-theoretic user model that is updated and corrected over time. It is designed to support the complex notion of shifting user interests and preferences by combining his long-term and short-term interests intuitively. The context in which the user seeks information is modeled by the agent through analyzing the user's interactions with the system to dynamically construct an ontology of concepts representing the user's information seeking context. User feedback is used to verify concepts prior to their commitment to a context knowledge model and to make requests to the target system to update the domain knowledge when needed.