In this paper we present a framework for identifying the test focus and test objectives based on the assumption that automatic information processing based on encoded meaning is the core of semantic e-government applications to be evaluated. Taking into account test strategies from software engineering and IT project management as well as different stakeholder perspectives, we identify possible test instruments. Several of these instruments have been applied in the Access-eGov project and we discuss the experience gathered in view of the newly developed framework in order to identify lessons learnt as well as to point to future research. The contribution of the paper is a portfolio of test strategies suggesting certain instruments to be applied from a systems view, agent view, and user view. We conclude that improving semantic e-government applications could be supported through applying a test-first approach, e.g. through providing an e-government test agent to be used in test labs or...