A comprehensive analysis of cyber attacks is important for better understanding of their nature and their origin. Providing a sufficient insight into such a vast amount of diverse (and sometimes seemingly unrelated) data is a task that is suitable neither for humans nor for fully automated algorithms alone. Not only a combination of the two approaches but also a continuous reasoning process that is capable of generating a sufficient knowledge base is indispensable for a better understanding of the events. Our research is focused on designing new exploratory methods and interactive visualizations in the context of network security. The knowledge generation loop is important for its ability to help analysts to refine the nature of the processes that continuously occur and to offer them a better insight into the network security related events. In this paper, we formulate the research questions that relate to the proposed solution.