Systems in general
and intelligent systems in particular, need to be able to explain their behaviour to their users or partners. Previously, a number of different user goals that explanations can support have been identified. Likewise, different kinds of explanations have been proposed. The problem remains how these abstract concepts can be made fruitful for the design of intelligent systems – they must be connected to software engineering methodologies. The work presented here builds on the concept of patterns and suggests using problem frames as a tool for requirements engineering. We further on propose to connect these problem frames with other design patterns as a tool supporting the implementation process.