To effectively support the user during the interactive IR process we need to capture the actual context, task or cognitive space of the user. 

Based on the logging of the information dialog of the user during the information search process we propose to use the implicit and explicit feedback give to either provide better query terms or to re-rank the result list accordingly.  

The context model about the user consists of several actions like entered query terms, viewed and  stored documents or associated journals and conferences. All information over time lead to a rich context representation which can be used to solve the actual task.