Much of our discussion has focussed on the interactive components of HAWK. However, in the problem domain of interest to DARPA there are many existing knowledge assets of great potential power. What prevents us from effectively utilitizing the knowledge in these assets is that the format is not appropriate for current knowledge processing technology; in particular, much of this information is captured in texts and databases. We will experiment with techniques for extracting this information and transforming it into usable knowledge representations.