Ramp-up of precision assembly lines is a cost-intensive and experience-driven task. Most of the time the knowledge how to effectively and efficiently setup an assembly line is intrinsic and is therefore neither shared nor reused by production experts. Almost no machine data is recorded until the correct functionality of the line is achieved and human problem solving tasks are not or poorly documented. In this paper a novel approach for structuring operator knowledge and combining it with machine-derived data by the use of semantic technologies is proposed. This enables human operators to express their experience in an easy to understand, machine readable way and makes it therefore accessible to other workers.