The incorporation and exploitation of background knowledge in KDD is essential for the effective discovery of useful patterns and the elimination of trivial results. We observe background knowledge as a combination of beliefs and interestingness measures. In conventional data mining, background knowledge refers to the preferences and properties of the population under observation. In applications analysing the interaction of persons with a system, we identify one additional type of background knowledge, namely about the strategies encountered in pursuit of the interaction objectives. We propose a framework for the modelling of this type of background knowledge and use a template-based mining language to exploit it during the data mining process. We apply our framework on Web usage mining for Web marketing applications.